In patients with cirrhosis, serum albumin determination should be carried out by immunonephelometry rather than by protein electrophoresis.

Objective Serum albumin is a key parameter for prognosis in cirrhosis. We compared levels of serum albumin determined by both protein electrophoresis and immunonephelometry, with special reference to the Child-Pugh classification. Design and methods One hundred and thirty-one patients, including 39 with cirrhosis, were included prospectively during 2 months. The aetiology of cirrhosis was mainly alcoholism (67%) and hepatitis C virus (HCV) (18%). Serum albumin was determined simultaneously by electrophoresis (Hydrasys® SEBIA following protein determination by the biuret reaction) and by immunonephelometry (BECKMAN® Nephelometer). Values were compared by non-parametric tests. Results For the whole population, electrophoretic and immunonephelometric values correlated (p = 0.85; P<0.0001), but electrophoresis significantly overestimated serum albumin by a median 1.6 g/l (P< 0.0001) with a large spread in values (range, -3.9 to 12.7). Median overestimation in cirrhosis was 2.6 g/l (P< 0.0001; range, -2.0 to 10.2) and 1.0 g/l (P< 0.0001; range, -3.9 to 12.7) in patients without cirrhosis (difference, P < 0.02). For 6/39 (15.4%) patients with cirrhosis, this overestimation led to an underestimation in the Child-Pugh classification. Conclusion In our experience, electrophoresis can lead to serum albumin values which are significantly different compared to those obtained by immunonephelometry. This discrepancy may lead to an incorrect Child-Pugh classification. Therefore, in the follow-up of cirrhotic patients, serum albumin should be determined by immunonephelometry.