Predictive value of systemic inflammatory response index for coronary artery disease and angiographic severity: a single-center retrospective study

Background Systemic inflammation plays a pivotal role in coronary artery disease (CAD). Systemic inflammatory response index (SIRI), a composite biomarker derived from neutrophil, monocyte, and lymphocyte counts, has been poorly studied for its association with angiographic CAD severity in East Asians. This study aimed to explore the relationship between SIRI and the presence and angiographic severity of CAD, and to develop a predictive model for risk stratification in a southeastern Chinese cohort. Methods We retrospectively enrolled adult inpatients who underwent coronary angiography in Wenzhou, China, between June 2023 and March 2024. CAD was defined as ≥50% stenosis in at least one major epicardial coronary artery. Baseline clinical, laboratory, echocardiographic, and angiographic data were collected. SIRI and other inflammation-based indices were calculated from admission blood counts. Multivariable logistic regression was used to identify independent CAD predictors, and the final CAD model was evaluated using analysis of receiver-operating characteristic curves, internal bootstrap validation, calibration assessment, decision curve analysis, added predictive value analysis, stability testing, and subgroup analyses. Results Among 394 patients, SIRI was higher in patients with CAD and increased across CAD phenotypes, with the highest values in the myocardial infarction groups. SIRI showed the highest AUC among the evaluated inflammatory indices for CAD discrimination. Albumin, left ventricular ejection fraction (LVEF), and SIRI remained independently associated with CAD. The albumin + LVEF + SIRI model showed moderate discrimination for angiographic CAD (AUC = 0.740, 95% CI: 0.682–0.797), improved the albumin + LVEF model (delta AUC = 0.045; DeLong p = 0.024), showed acceptable internal calibration, and remained stable when SIRI was modeled as untransformed, log 10 -transformed, or quartile-based values. A low-costnomogram (AUC = 0.74) showed favorable internal calibration and parallel performance for angiographic severity (AUC = 0.719). Conclusions SIRI was independently associated with angiographic CAD and disease severity in this cohort. A predictive model integrating SIRI, albumin, and LVEF showed moderate discrimination and should be considered a preliminary, internally validated, hypothesis-generating risk-stratification framework rather than a clinically deployable decision-making tool.