Each year, millions of patients worldwide undergo non-cardiac surgery (NCS), with over 18% of cases involving individuals with stable coronary artery disease (SCAD) (1). Among SCAD patients undergoing NCS, 5.7%–10.0% experience major adverse cardiovascular events (MACE) during the perioperative period (2–4), a rate significantly higher than the 2.5%–3.0% observed in the general surgical population (5). MACE—including cardiac arrest, myocardial infarction (MI), heart failure (HF), and stroke (6)—represents a major contributor to perioperative morbidity and mortality, imposing a considerable social and economic burden. Therefore, accurate prediction of perioperative MACE risk is crucial. It enables clinicians to comprehensively assess surgical risks, tailor perioperative management plans, facilitate informed shared decision-making with patients, and optimize the allocation of healthcare resources.

The development of postoperative cardiovascular events is influenced by multiple factors, among which inflammation has recently emerged as a potential driving mechanism. Inflammatory responses contribute to cardiovascular injury through various pathways. Chronic inflammation can impair endothelial function and promote the formation of atherosclerotic plaques. Inflammatory cytokines such as interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) enhance platelet aggregation and increase plaque vulnerability (7–9). Additionally, oxidative stress-induced inflammatory cascades may directly damage cardiomyocytes. Systemic inflammation is closely associated with the progression of atherosclerosis, alterations in hemostasis, the presence and severity of coronary artery disease (CAD), and the development of acute coronary thrombosis (10, 11). Collectively, these mechanisms promote plaque destabilization, thrombogenesis, and microvascular dysfunction, ultimately leading to MACE. Recent studies have indicated that surgical procedures significantly exacerbate inflammatory responses. Surgical trauma elicits a physiological stress response—including heightened sympathetic nervous system activity, catecholamine release, and hemodynamic fluctuations—that not only increases myocardial oxygen demand but also triggers systemic inflammation characterized by the release of proinflammatory cytokines and activation of immune cells. These processes may impair endothelial function, destabilize plaques, and induce microvascular dysfunction, thereby accelerating myocardial injury (12). Thus, evaluating the preoperative inflammatory status is critical for predicting perioperative cardiovascular risk in patients with SCAD.

Inflammatory biomarkers provide a quantitative means of assessing systemic inflammation, and recent studies have shown that inflammatory biomarkers are useful prognostic tools for cardiovascular disease and that their integration into risk stratification models may improve their predictive accuracy (13, 14). Several inflammation-based indicators—such as high-sensitivity C-reactive protein (hs-CRP), the systemic immune-inflammation index (SII), and the advanced lung cancer inflammation index (ALI)—have been shown to correlate with prognosis in patients with acute coronary syndrome (ACS) (15–18). Recently, the NPAR, a novel composite biomarker that integrates neutrophil percentage and serum albumin level, has been proposed as a sensitive indicator of persistent inflammation associated with cardiovascular injury (19). Owing to its cost-effectiveness and accessibility, NPAR has demonstrated prognostic value in patients with cancer and myocardial infarction (20). However, no studies to date have evaluated the predictive value of NPAR for perioperative cardiovascular events in patients with SCAD. Therefore, this study aims to investigate the association between preoperative NPAR and the incidence of MACE in patients with SCAD undergoing NCS, and to explore the prognostic utility of NPAR in this clinical setting.

Figure 3 illustrates the ROC curves and corresponding AUC values for predicting perioperative MACEs. The results show that NPAR achieved an AUC of 0.721, outperforming other markers such as albumin (AUC: 0.699) and neutrophil count (AUC: 0.604). This suggests that NPAR may serve as a more effective predictor of perioperative MACE. As shown in Table 4 and Figure 4, incorporating NPAR into the predictive model significantly improved its performance. Specifically, the addition of NPAR to the RCRI model increased the AUC from 0.679– 0.755, with a statistically significant difference confirmed by DeLong's test (Supplementary Table S4). This improvement translates into one additional high-risk patient being correctly reclassified for every 13 patients evaluated.

The C-statistic, a key indicator of model discrimination, also improved when NPAR was added as a continuous variable to the RCRI model, demonstrating enhanced predictive power both statistically and clinically (Table 5). Model fit was also significantly improved, with a continuous NRI of 0.599 (95% CI: 0.311–0.802, p < 0.001) and an IDI of 0.035 (95% CI: 0.013–0.057, p = 0.002), underscoring the practical utility of these enhancements. In order to better apply NPAR in clinical practice, we built a nomogram (Figure 5) that combines NPAR and RCRI model to predict perioperative adverse cardiovascular events in patients with stable coronary heart disease, so that it can be used by clinicians. The calibration curve (Supplementary Figure S2) closely aligned with the reference line, indicating strong model calibration and predictive performance. Additionally, DCA (Supplementary Figure S3) demonstrated a meaningful net benefit, highlighting the model's potential to support clinical decision-making and real-world application.

Given the higher perioperative MACE risk in patients with SCAD compared to the general population, preoperative cardiovascular risk assessment in this high-risk group is critical for optimizing risk reduction strategies and improving surgical outcomes (30). This study investigated the association and predictive value of the NPAR in assessing perioperative MACEs in patients with stable coronary artery disease undergoing NCS. Our results demonstrated that elevated preoperative NPAR was significantly associated with an increased risk of perioperative MACEs. A J-shaped relationship was observed between NPAR and MACEs, with the risk rising notably when NPAR exceeded 13.7. Moreover, mediation analysis suggested that decreased eGFR may be one of the underlying mediators. ROC analysis and machine learning models further indicated that incorporating NPAR improved the predictive accuracy for perioperative MACEs. We have also constructed an effective perioperative predictive nomogram, and the perioperative risk probability of MACEs in patients with SCAD can be estimated by calculating scores. These findings suggest that NPAR may serve as a key predictor of adverse cardiovascular outcomes during the perioperative period.

As key components of the inflammatory response, neutrophils not only contribute to the development of coronary atherosclerosis but also play a critical role in atherothrombotic events, adverse remodeling, mortality, and the occurrence of MACEs following ACS (31, 32). Albumin has traditionally been regarded as a marker of nutritional status. It is well established that low serum albumin levels reflect the severity of inflammation and contribute to the progression of cardiovascular diseases (CVDs) (33). Furthermore, hypoalbuminemia has been associated with increased mortality and poor prognosis in patients with myocardial infarction (34). Physiological stress responses — including heightened sympathetic nervous system activity, catecholamine release, and hemodynamic alterations—can be triggered by surgery. Surgical trauma also induces a systemic inflammatory response, which may increase myocardial oxygen demand and accelerate myocardial injury, particularly in susceptible individuals (12). As a combination of two inflammation-related markers, NPAR reflects both nutritional status and the degree of inflammation. Some studies have suggested that NPAR may provide a more comprehensive assessment of inflammatory responses (35, 36). Mechanistically, perioperative sterile inflammation activates neutrophils and promotes the formation of neutrophil extracellular traps (NETs), which disrupt the endothelial glycocalyx, reduce nitric oxide bioavailability, and impair microcirculation. At the same time, inflammation lowers serum albumin, diminishing its antioxidant and endothelial-protective effects. By combining increased neutrophil percentage with decreased albumin into a single ratio, NPAR may better capture this inflammation endothelium microcirculation pathway, which could explain its stronger prognostic value compared with either marker alone.

Activation of neutrophils and the release of inflammatory mediators such as TNF-α and interleukin-6 (IL-6) contribute to vascular endothelial injury, exacerbate vascular dysfunction, and promote atherosclerosis (37). This inflammatory cascade leads to elevated NPAR levels. In addition, oxidative stress, through the generation of reactive oxygen species (ROS), further damages cellular structures, worsens vascular health, and perpetuates the vicious cycle of inflammation. Chronic inflammation has also been shown to impair albumin synthesis, thereby further compromising patient health (38). In recent years, studies have shown that NPAR is an inflammation-based prognostic predictor for various cardiovascular diseases, including atrial fibrillation, chronic heart failure, acute ST-elevation myocardial infarction (STEMI), and cardiogenic shock (31, 35, 36, 39, 40). For example, a retrospective study by Wang et al. involving 622 patients with heart failure found that patients with higher NPAR levels had significantly increased mortality at 90 days (HR, 95% CI: 2.21, 1.01–4.86), 1 year (HR, 2.13, 95% CI: 1.30–3.49), and 2 years (HR, 2.06, 95% CI: 1.37–3.09), compared with those with lower NPAR levels (P for trend < 0.05). They also demonstrated that NPAR not only correlated with all-cause mortality in patients with chronic heart failure but also provided better prognostic value than either albumin or neutrophil percentage alone (35). An analysis based on the MIMIC-III database showed that among 3,106 critically ill patients diagnosed with coronary artery disease, patients in the highest NPAR quartile had significantly higher all-cause mortality at 30 days (HR, 1.924, 95% CI: 1.471–2.516), 90 days (HR, 2.053, 95% CI: 1.646–2.560), and 365 days (HR, 2.063, 95% CI: 1.717–2.480) compared to those in the lowest quartile (34). In a study by Cui et al., NPAR was identified as an independent predictor of in-hospital mortality among STEMI patients (41). Additionally, two studies involving patients undergoing peritoneal dialysis consistently found that higher NPAR levels were independently associated with increased risks of all-cause and cardiovascular mortality (42, 43). Among critically ill patients, elevated NPAR levels at admission were also independently associated with higher all-cause mortality and prolonged hospital stay (44). Moreover, a previous cross-sectional study demonstrated that elevated NPAR levels were associated with a higher prevalence of chronic kidney disease (CKD), lower eGFR, and increased proteinuria, with these associations consistently observed across different population settings (27). Given that impaired renal function is an established risk factor for MACEs, these findings suggest that eGFR may serve as a potential mediator in this relationship. Our study yielded similar results, supporting this hypothesis. Furthermore, although eGFR only accounted for 8.4% of the NPAR–MACE association—indicating a limited role as a single mediator—the perioperative inflammation–endothelium–microcirculation axis likely also involves cytokines such as IL-6 and TNF-α, which are not routinely measured, as well as surgical trauma intensity and anesthesia modality. Multicenter prospective studies that systematically collect these variables are therefore warranted to further dissect the multifactorial mechanisms by which elevated NPAR drives perioperative cardiovascular events.

Currently, the RCRI is the most commonly used model for predicting perioperative MACE risk. However, recent studies have indicated that RCRI has limited discriminatory ability (45), particularly among patients with known SCAD (46). Incorporating NPAR into the RCRI model may enhance its predictive performance for MACEs. Although RCRI is widely used in NCS, it often lacks the precision required for individualized risk assessment (47, 48). For example, data from the VISION cohort showed that among patients aged 45 years and older, one in twelve individuals without traditional RCRI risk factors still experienced cardiac complications following major non-cardiac surgery (49). Many studies have explored the use of biomarkers to improve the predictive ability of RCRI in the non-cardiac surgical setting. The value of brain natriuretic peptide (BNP) in preoperative risk assessment and cardiac troponin I (cTnI) in detecting prognostically relevant myocardial ischemia has been well established (50). Compared to traditional cardiac biomarkers, NPAR may be a more suitable preoperative screening tool due to its cost-effectiveness and broad applicability. NPAR offers a promising alternative for personalized risk assessment, potentially identifying high-risk patients who might otherwise be overlooked. The combination of the XGBoost classifier and the Boruta algorithm has been widely applied across various fields, including medicine, computer science, molecular biology, and economics, demonstrating superior overall model performance compared to other approaches (51–54). In medicine, this approach has been used not only for disease diagnosis and prognosis but also for biomarker discovery and risk prediction (51, 54). Both the Boruta algorithm and SHAP analysis in our study identified NPAR as an important predictor of perioperative MACE risk. To our knowledge, this is the first study to investigate the association between NPAR and perioperative MACEs in patients with SCAD, as well as to evaluate the prognostic value of integrating NPAR into the RCRI model. NPAR may serve as a reliable, inexpensive, and easily accessible biomarker for assessing the risk of perioperative MACEs in SCAD patients. By utilizing NPAR for MACE risk assessment, our study provides a practical tool to help clinicians identify high-risk patients early, guiding preventive strategies and timely interventions in clinical practice. Routine monitoring of inflammatory markers such as neutrophil counts and albumin levels could facilitate early detection of high-risk individuals and timely intervention. Furthermore, exploring the molecular mechanisms underlying the association between NPAR and perioperative MACEs in SCAD patients may offer insights into potential therapeutic targets. This study was conducted at a single center, limiting the ability to fully eliminate biases related to patient selection, regional healthcare differences, and variations in follow-up management. As a result, the generalizability of the findings is restricted. Additionally, the low number and incidence of outcome events may affect the stability of the multivariable models and the generalization performance of the predictive models. Although the machine learning models demonstrated good performance in internal cross-validation, they have yet to be validated in independent external cohorts or prospective studies. Further external validation is needed to confirm the robustness and clinical applicability of the models. Despite constructing multiple adjusted models and controlling for numerous covariates, inherent unmeasured confounding factors associated with the retrospective design may still impact the authenticity of the results. Moreover, while our mediation analysis identified eGFR as a partial mediator in the association between NPAR and perioperative MACE, the proportion mediated was relatively modest (8.4%), suggesting that other unmeasured pathophysiological pathways—such as those involving inflammatory cytokines (e.g., IL-6, TNF-α), intensity of surgical stress, or specific anesthesia techniques—may contribute more substantially to this relationship. The absence of systematically collected data on these potential mediators represents an important limitation of our study. In addition, the relatively short follow-up period may not have fully captured the long-term prognostic impact of NPAR, and the presence of some missing data could have introduced potential bias despite our efforts at statistical adjustment. In summary, before incorporating NPAR into clinical decision-making, multicenter prospective studies are required, with routine postoperative monitoring of inflammatory and myocardial injury markers and longer follow-up periods, to externally validate the models and further assess their applicability across different populations and healthcare settings.

This study identified the NPAR as an independent predictor of MACE during the perioperative period in patients with stable coronary artery disease undergoing NCS. Elevated preoperative NPAR levels were significantly associated with an increased risk of MACE, with eGFR potentially acting as a mediator in this association. Integrating NPAR into established risk assessment models—particularly the RCRI—substantially improved predictive performance. Additionally, the development of a nomogram allows for individualized risk estimation, facilitating clinical decision-making. The prognostic value of NPAR was further validated through machine learning algorithms, reinforcing its potential role in perioperative risk stratification. These findings provide a strong foundation for future prospective cohort studies and mechanistic investigations to confirm the clinical utility and pathophysiological relevance of NPAR in this setting.