Comparison of acute kidney injury classifications in patients undergoing transcatheter aortic valve implantation: Predictors and long-term outcomes

Background Acute kidney injury (AKI) was demonstrated to adversely affect outcome in patients undergoing transcatheter aortic valve implantation (TAVI). We compared predictors for AKI and associated outcomes according to various definitions among patients undergoing TAVI in a tertiary medical center. Methods Two-hundred and seventeen TAVI patients were evaluated for the occurrence of AKI according to Kidney Disease Improving Global Outcomes (KDIGO)/Valve Academic Research Consortium (VARC-2) and Risk Injury Failure Loss End-Stage (RIFLE) definitions. Multivariate analysis was conducted to assess predictors of AKI. Cox hazard ratio was used to evaluate long-term mortality in this patient population. Results AKI occurred in 23 and 21% of patients (n = 49, n = 46) according to KDIGO/VARC-2 and RIFLE definitions, respectively, with an approximate 10% of disagreement between both systems. Predictors of AKI according to KDIGO/VARC-2 were chronic obstructive pulmonary disease (COPD; OR = 2.66, P = 0.01), PVD (OR = 3.45, P = 0.02) and a lower baseline eGFR (OR = 1.03 per 1 mL/min/1.73 m² decrease, P = 0.02). While BMI (OR = 1.12, P = 0.01), prior ischemic heart disease (OR = 2.35, P = 0.04) and COPD (OR = 2.18, P = 0.04) were associated with AKI as defined by the RIFLE definition. AKI defined by either classification was independently associated with long-term mortality (HR = 1.63, for the KDIGO/VARC-2 definition and HR = 1.60 for RIFLE definition, P = 0.04 for both models), with borderline superiority of the KDIGO/VARC-2 classification. Conclusions Different clinical characteristics predict the occurrence of AKI after TAVI when RIFLE and KDIGO/VARC-2 classifications are used. Both classification systems of AKI identify patients with increased risk for long-term mortality, with superiority of the KDIGO/VARC-2 definition, which should be used for AKI grading.