Eur Heart J Qual Care Clin Outcomes. 2026 Jul 28:qcag122. doi: 10.1093/ehjqcco/qcag122. Online ahead of print.
ABSTRACT
BACKGROUND: Accurate risk stratification for patients with three-vessel coronary artery disease (3VD) undergoing percutaneous coronary intervention (PCI) remains important in contemporary practice. SYNTAX-based mortality prediction models require reassessment in modern PCI populations.
METHODS: This post-hoc analysis of the Multivessel TALENT trial evaluated the core and extended logistic clinical SYNTAX Score (LCSS) for predicting 1-year all-cause mortality. Discrimination was assessed using the area under the receiver-operating characteristic curve (AUC), and calibration using calibration intercept, calibration slope, graphical calibration, and the E-statistic. Prediction scores were calculated within each of 20 imputed datasets, with model performance evaluated within each imputation and summarised across imputations. Intercept-and-slope recalibration and decision curve analysis were also performed.
RESULTS: At 1 year, 46 (3.0%) of 1,548 enrolled patients had died. The pooled AUCs were 0.716 for the LCSS core model and 0.744 for the extended model, compared with 0.629 for the anatomical SYNTAX Score and 0.632 for the functional SYNTAX Score. LCSS models systematically overestimated absolute risk, although observed mortality increased across predicted-risk quintiles. Decision curve analysis showed a positive net benefit for the original and recalibrated LCSS models across threshold probabilities of 1% to 10%, with numerically higher net benefit for the recalibrated extended model across much of the evaluated threshold range.
CONCLUSION: In contemporary PCI for 3VD, the LCSS showed moderate discrimination for 1-year all-cause mortality while overestimating its absolute risk. Recalibration improved agreement with observed risk in this cohort and may inform future validation and model refinement.
PMID:42509586 | DOI:10.1093/ehjqcco/qcag122
