Estimated glomerular filtration rate by formulas in patients with cirrhosis: An unreliable procedure.

Abstract

In cirrhosis, the reliability of formulas that estimate renal function, either those specifically developed in this population or the classic equations, has not been properly quantified. We studied the agreement between estimated (eGFR) and measured glomerular filtration rate (mGFR) in cirrhosis. Renal function was estimated with 56 formulas including specific equations: Glomerular Filtration Rate Assessment in Liver Disease (GRAIL), Royal Free Hospital Cirrhosis (RFHC) and Mindikoglu-eGFR, and measured with a gold standard procedure; plasma clearance of iohexol using dried blood spots sampling in a group of cirrhotics. The agreement eGFR-mGFR was evaluated with specific tests: total deviation index (TDI), concordance correlation coefficient (CCC) and coverage probability (CP). We defined acceptable agreement as values: TDI<10%, CCC≥0.9 and CP>90%. A total of 146 patients (age 65±9years, 81% male) were evaluated; 61 (42%) Child A, 67 (46%) Child B and 18 (12%) Child C. Median MELD-Na was 14 (9-15). The agreement between eGFR and mGFR was poor: TDI averaged was of 73% (90% of the estimations ranged from ±73% of mGFR); CCC averaged was 0.7 indicating low concordance and CP averaged 22% indicating that 78% of the estimations have an error>10%. Specific formulas showed also poor agreement: TDI was 82%, 70% and 37% for the GRAIL, RFHC and Mindikoglu equations, respectively. Overall, formulas poorly estimated renal function in cirrhotic patients. Specific formulas designed for cirrhosis did not outperform classic equations. eGFR must be considered with caution in cirrhotic patients.