Risk Factors and Predictive Model for In‐Stent Restenosis After PCI in Coronary Bifurcation Lesions: A Retrospective Study
Zhicong Gu, Erkun Xing, Xiu Jin, Youqi Zhu, Lei Dang, Ming BaiABSTRACT
Background
In‐stent restenosis (ISR) after percutaneous coronary intervention (PCI) in true bifurcation coronary lesions can lead to repeat revascularization and adverse outcomes.
Aims
To identify independent risk factors for ISR in patients with coronary bifurcation lesions undergoing PCI and to develop a predictive model for ISR‐free survival.
Methods
Patients with coronary artery disease and Medina coronary bifurcation lesions was analyzed. Baseline characteristics, laboratory values, bifurcation anatomy, and PCI details were collected. Cox proportional hazards models were used to identify independent predictors of ISR and its performance was evaluated Model interpretability was explored with Shapley additive explanations (SHAP), and a nomogram was constructed for clinical use.
Results
Univariate analysis showed lesion/procedural factors were strongly predictive. Multivariable Cox analysis identified use of a drug‐coated balloon in the side branch and side‐branch stenting with a drug‐eluting stent as the top independent predictors of ISR. Lesion location in the left circumflex was associated with higher ISR risk The Cox model Calibration plots showed close agreement between predicted and observed ISR probabilities. A SHAP analysis confirmed the direction and relative importance of each predictor, enhancing the model ' s interpretability.
Conclusions
In true bifurcation PCI, Side‐branch interventions are associated with higher ISR risk, whereas a simple single‐stent approach confers lower risk. The developed Cox‐based nomogram demonstrated robust ability to predict ISR, which can aid clinicians in risk stratification and personalized management. Early identification of high‐risk patients may enable closer surveillance or preventive measures to improve long‐term outcomes after bifurcation PCI.