Preoperative high-sensitivity C-reactive protein-to-albumin ratio predicts arteriovenous fistula failure in hemodialysis patients: a retrospective cohort study with landmark analysis

Abstract

Background The arteriovenous fistula (AVF) is the preferred vascular access for hemodialysis, yet up to 50% of AVFs fail within the first year due to stenosis or thrombosis. The high-sensitivity C-reactive protein to albumin ratio (HRR) integrates inflammation and nutritional status, but its association with AVF failure has not been systematically evaluated. Methods This retrospective cohort study included 279 patients who underwent first-time AVF creation. HRR was calculated as hsCRP (mg/L) divided by albumin (g/L). The primary outcome was AVF failure (stenosis ≥50% or thrombosis requiring intervention). Multivariate Cox regression adjusted for age, sex, BMI, diabetes, hypertension, smoking, and vessel diameter. Model performance was assessed using time-dependent ROC, calibration curve, and decision curve analysis (DCA). Results During a median follow-up of 50 months, 141 patients (50.5%) developed AVF failure.In multivariate analysis, higher HRR was independently associated with increased AVF failure risk (HR per unit increase = 1.65, 95% CI: 1.33–2.05, P < 0.001), independent of vessel diameter (HR = 0.21, 95% CI: 0.08–0.57, P = 0.002). A landmark analysis stratified at 24 months demonstrated that the predictive effect of HRR was directionally consistent across early (HR = 1.60, 95% CI: 1.27–2.02) and late (HR = 1.47, 95% CI: 0.66–3.29) follow-up periods (P for interaction = 0.766), although the late-phase estimate was imprecise due to the reduced sample size. The model demonstrated moderate discrimination at 24 months (AUC = 0.608, 95% CI: 0.519–0.701) and good calibration. DCA confirmed that incorporating HRR provided positive net clinical benefit across clinically relevant threshold probabilities. The proportional hazards assumption was satisfied (global P = 0.97). Sensitivity analysis using multiple imputation yielded consistent results. Conclusions Preoperative HRR is an independent predictor of AVF failure. Although the model demonstrated weak to moderate discrimination (24-month AUC = 0.608), decision curve analysis confirmed a positive net clinical benefit, supporting its potential utility in preoperative risk stratification. This simple, widely available laboratory-based marker may assist clinicians in identifying high-risk patients and individualizing surveillance intensity, though external validation is warranted.