Creatinine-based glomerular filtration rate estimation in patients with liver disease: the new Chronic Kidney Disease Epidemiology Collaboration equation is not better.

Objectives: Limitations of serum creatinine in patients with an impaired liver function are well known. The commonly used modification of diet in renal disease (MDRD) equation has a low diagnostic performance to approximate kidney function in patients after liver transplantation (LT) and patients with liver cirrhosis (LC). The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula has been shown to provide a more accurate estimation of kidney function in patients with chronic kidney disease, but studies in patients with liver disease are lacking.

Methods: We evaluated the diagnostic performance of CKD-EPI in comparison with the re-expressed MDRD formula in patients after LT (group 1; n=59) and in patients suffering from LC (group 2; n=44). GFR was measured by 99mTc-diethylenetriamine penta-acetic acid (group 1) and inulin clearance (group 2). Bias, precision, and accuracy as compared with the measured GFR were determined.

Results: The measured mean GFR (95% confidence interval) was 52.3 ml/min/1.73 m2 (47.7; 56.9; group 1) and 35.3 ml/min/1.73 m2 (29.12; 41.3; group 2), respectively. In transplanted patients GFR estimation by CKD-EPI and MDRD did not significantly differ with respect to bias (9.7 vs. 4.3 ml/min/1.73 m2), precision (16.9 vs. 15.5 ml/min/1.73 m2) and accuracy (64.4 vs. 69.5% within 30% of 'true GFR'). In patients with LC, both formulae showed a very high bias (42.5 vs. 40.1 ml/min/1.73 m2), a very low precision (20.7 vs. 25.7 ml/min/1.73 m2) and accuracy (6.8 within 30% of the measured GFR in both groups).

Conclusion: The CKD-EPI equation does not improve the creatinine-based GFR estimation in patients after LT. In patients with LC, both equations should not be applied as they extremely overestimate GFR.