Development and validation of a predictive model for side branch flow impairment following stent implantation in patients with non-left main coronary bifurcation lesions: a retrospective analysis

Abstract Objectives The purpose of this study was to develop and validate a predictive model to assessing the risk of side branch flow impairment (SBFI) following stent implantation in patients with non-left main coronary bifurcation lesions (CBLs). This model aims to provide preprocedural risk stratification and inform the selection of interventional strategies. Background Coronary artery bifurcation lesions constitute a particularly complex subtype of coronary artery disease that is frequently encountered in practice. Compared with non-bifurcation lesions, they are associated with greater procedural complexity and risk of procedure-related complications. Patients and methods Data from 830 patients with CBL who underwent percutaneous coronary intervention (PCI) in the Affiliated Hospital of Chengde Medical University from January 2022 to December 2023 were retrospectively collected. The least absolute shrinkage and selection operator regression methods were used to screen variables, and multivariate logistic regression was used to establish a predictive model. A nomogram was built based on these factors and internally verified using the bootstrap resampling method. The C-statistic was used to verify and evaluate the discriminative ability of the model; the calibration curve was drawn, and the decision curve analysis (DCA) was performed to evaluate the calibration degree, clinical net benefit, and practicability of the model. The primary endpoint was SBFI, defined as a transient or persistent reduction in thrombolysis in myocardial infarction (TIMI) flow grade in a branch vessel following stent implantation in a major non-left main coronary artery. Results A nomogram was constructed using the selected predictors of SBFI, which included age, plaque location ipsilateral to the SB, TIMI flow grade before main vessel (MV) stenting, and N-terminal pro-brain natriuretic peptide (NT-proBNP). The discriminatory ability of the model, as assessed by the area under the curve (AUC), was 0.651. The robustness of the model was evaluated through internal validation with 1000 bootstrap replicates, resulting in a corrected AUC of 0.641. The calibration curve, evaluated by the Hosmer–Lemeshow test, showed good agreement between predictions and observations (χ 2 = 5.765, P = 0.674). Finally, DCA indicated that the model provided clinical net benefit over a threshold probability range of 0.19–0.45. Conclusion We developed and validated a predictive model for SBFI after PCI in non-left main bifurcation lesions. The model exhibited modest discriminative ability, calibration, and a positive net benefit on DCA, suggesting that it may serve as a valuable risk stratification tool in clinical practice.