Introduction Although serum creatinine is a well-recognised marker of prognosis in cirrhosis, it is only an indirect marker of renal function as it is affected by various extra-renal factors. The measurement of glomerular filtration rate (GFR) by the plasma clearance Cr-EDTA is an acceptable substitute of the gold standard of inulin clearance. We assessed the correlations of serum creatinine with GFR measured by Cr-EDTA in patients with cirrhosis in comparison with patients with renal disease.

Method We analysed data from 298 consecutive patients who underwent GFR assessment by Cr-EDTA as part of their liver transplant work-up. We collected similar data on 187 consecutive non-cirrhotic patients who attended the renal outpatient clinic. GFR was assessed by bolus infusion of Cr-EDTA and single or serial serum measurements after 2, 4, 6 and 24 h. Spearman test was used to correlate serum creatinine and GFR in renal and liver patients. The significance of the difference between the correlations from the two groups was calculated by transforming the Spearman's r to Fischer's z-score, estimating the SE of difference between the two correlations and finally dividing the differences between the two z-scores by the SE. If the result was 1.96 or higher, then the difference in the correlation was considered significant in the 0.05 level.

Results Serum creatinine significantly and inversely correlated with GFR in patients with cirrhosis (r=−0.702, p<0.001) and renal disease (r=−0.856, p<0.001), however the difference of the correlation was significant between patients with renal disease and patients with cirrhosis (p<0.05). When analysis was performed according to gender, there were significant correlations of serum creatinine and GFR in patients with cirrhosis (males r=−0.806 and females r=−0.699) and renal disease (males r=−0.877 and females r=−0.890). Moreover, the difference of the correlation was again significant among male and female patients with renal disease and cirrhosis and notably in male compared to female patients with cirrhosis (p<0.05). Therefore, for a given GFR, patients with cirrhosis have lower serum creatinine values than patients with renal disease. Moreover, female patients with cirrhosis have lower serum creatinine values than male patients with the same GFR.

Conclusion Serum creatinine underestimates renal function in patients with cirrhosis compared to patients with renal disease. Serum creatinine cut-offs used to define renal failure in the general population are not applicable to patients with cirrhosis and should be re-evaluated as they systematically underestimate renal function.