Systemic Inflammation Response Index (SIRI) and Contrast-Associated AKI in Neurointerventional/Peri-procedural Setting

Background Contrast-associated acute kidney injury (CA-AKI) remains clinically relevant in neurointerventional/peri-procedural pathways, yet bedside-feasible risk markers are limited. We evaluated whether the systemic inflammation response index (SIRI) independently predicts CA-AKI and improves model performance. Methods We performed a single-center retrospective cohort study of consecutive patients undergoing neurointerventional or time-sensitive neurovascular pathways with iodinated contrast exposure (typically CTA/CTP and/or DSA with endovascular treatment) during the study period. CA-AKI was defined as an absolute rise in serum creatinine ≥ 26.5 µmol/L from the pre-procedural value to the highest post-procedural value measured within [48–72 hours / 7 days] after contrast exposure during the index hospitalization (using the first available post-procedural creatinine and subsequent inpatient measurements when available).SIRI was calculated as neutrophils × monocytes / lymphocytes and modeled as ln(SIRI) per 1 SD. We fitted prespecified logistic models (Model 1: SIRI; Model 2: age + sex+SIRI; Model 3: age + sex+atrial fibrillation+hypertension+diabetes+baseline creatinine+SIRI). Discrimination (AUC), bootstrap optimism-corrected calibration (B = 600), and decision curve analysis (DCA; threshold probability 0.05–0.40) were assessed. Results CA-AKI incidence was 17.9% (40/223). Median SIRI was higher in CA-AKI (3.13 [1.98–4.29]) than Non-AKI (1.54 [0.99–2.78], P < 0.001). In Model 3, ln(SIRI) per SD remained independently associated with CA-AKI (aOR 1.87, 95% CI 1.27–2.75; P = 0.002); atrial fibrillation (OR 3.96) and hypertension (OR 2.96) were also independent predictors. A parsimonious clinical model (age+atrial fibrillation+hypertension) yielded AUC 0.675, which improved to 0.732 after adding SIRI (ΔAUC + 0.057). The optimal SIRI cutoff was 1.80 (sensitivity 0.800, specificity 0.617). The combined model showed acceptable optimism-corrected calibration (slope ≈ 0.88, intercept ≈ − 0.16) and higher net benefit than the clinical model across key thresholds (notably 0.05–0.26) on DCA. Conclusions Conclusions: Admission SIRI independently predicted CA-AKI and improved discrimination and decision-analytic net benefit when added to pragmatic clinical variables; external validation is required before clinical implementation.