Comparison of prognostic efficacy of acute kidney injury criteria in patients with liver cirrhosis

Abstract

Background/Aims: Development of acute kidney injury (AKI) is closely associated with mortality in patients with liver cirrhosis. Recently, several new definitions of AKI were published. This study was performed to compare the efficacy of several definitions of AKI for predicting prognosis in cirrhotic patients.

Methods: Cirrhotic patients who hospitalized to our hospital were enrolled. Patients with hepatocellular carcinoma and parenchymal kidney disease were excluded. AKI was defined by con-ventional, RIFLE, and KDIGO criteria.

Results: A total of 696 cirrhotic patients were enrolled. Age was 54.1±10.7 years and 526 patients (75.6%) were men. Child-Pugh and MELD scores were 8.7±2.4 and 15.5±6.9, respectively. During follow-up, 155 patients died. Six and 12 months mortality rates were 10.7% and 14.2%, respectively. Among all patients, 22 (3.2%), 29 (4.2%), and 52 (7.5%) patients fulfilled the criteria of AKI of conventional, RIFLE, and KDIGO definitions, respectively. Survival time differed significantly between patients without and with AKI according to the conventional (92.3±2.1 vs 18.7±6.2 months, P<0.001), RIFLE (92.3±2.2 vs 42.8±11.2 months, P<0.001), and KDIGO (94.0±2.2 vs 44.0±8.3 months, P<0.001) criteria. Both conventional and KDIGO criteria were fulfilled in 22 patients (3.2%), while 30 patients (4.3%) fulfilled only KDIGO criteria. There was a trend of worse prognosis in patients who fulfilled both conventional and KDIGO criteria than those who only fulfilled KDIGO criteria (18.7±6.2 vs 53.3±11.1 months, P=0.051). Both RIFLE and KDIGO criteria were fulfilled in 29 patients (4.2%), while 23 patients (3.3%) fulfilled only KDIGO criteria. Survival did not differ between patients who fulfilled both RIFLE and KDIGO criteria and those who fulfilled only KDIGO criteria (42.8±11.2 vs 30.1±7.7 months, P=0.106).

Conclusion: KDIGO criteria for AKI predicted survival in cirrhotic patients more accurately compared to conventional and RIFLE criteria. Our results suggest that even subtle changes in the serum creatinine level requires close attention in these patients.