Comparative predictive performance of different eGFR equations for 1-year major adverse cardiovascular events after percutaneous coronary intervention in patients with acute coronary syndrome.

BACKGROUND: Renal dysfunction is an important determinant of cardiovascular prognosis in patients with coronary artery disease. Estimated glomerular filtration rate (eGFR) is widely used to assess kidney function, but different equations may yield varying risk stratification results. This study aimed to compare the predictive performance of creatinine-based and cystatin C-based eGFR equations for 1-year major adverse cardiovascular events (MACE) in patients undergoing percutaneous coronary intervention (PCI). METHODS: This retrospective study included 560 patients who underwent PCI for acute coronary syndrome. Baseline clinical, laboratory, and echocardiographic variables were collected. Renal function was assessed using both creatinine-based and cystatin C-based eGFR equations. Logistic regression with backward selection based on the Akaike Information Criterion (AIC) was performed to identify independent predictors of 1-year MACE. In addition, multiple machine learning models were developed to evaluate predictive performance, and Shapley Additive Explanations (SHAP) analysis was applied to assess feature importance. RESULTS: During the 1-year follow-up, MACE occurred in 72 patients, comprising 71 repeat PCIs and one cardiac death. Equations relying on cystatin C detected a larger share of patients with impaired renal function compared with creatinine-based formulas. However, moderate-to-severe renal dysfunction defined by creatinine-based eGFR was significantly associated with an increased risk of 1-year MACE, although it was not retained as an independent predictor after AIC-based multivariable adjustment. Machine learning models demonstrated modest predictive performance overall, with logistic regression achieving the highest discriminative ability. SHAP analysis indicated that cardiovascular functional parameters and metabolic markers were the major contributors to risk prediction, while renal function indicators provided complementary prognostic information. CONCLUSIONS: Creatinine-based eGFR showed a stronger observed association with 1-year MACE after PCI than cystatin C-based eGFR in this single-center cohort; these findings do not establish clinical superiority of either equation and require external validation. Renal function assessment remains an important component of cardiovascular risk stratification, and traditional regression models performed comparably to machine learning approaches in moderately sized clinical datasets.