We read with great interest the recent work by Petta et al. suggesting serum levels of retinol-binding protein 4 (RBP4) as a marker for the degree of steatosis in patients infected with hepatitis C virus (HCV).1 RBP4 has provoked exceptional attention because it has been linked to the pathogenesis of insulin resistance in mice and humans.2, 3 Petta et al. now reported an association between elevated serum RBP4 and hepatic steatosis as well as insulin resistance in patients with nonalcoholic steatohepatitis (NASH), and serum RBP4 appeared to be a strong independent predictor of hepatic steatosis in nondiabetic, nonobese patients chronically infected with HCV genotype 1.1
However, from our own4 and other studies,5 we would like to stress important potential confounding factors that have neither been evaluated nor discussed by the authors and might greatly hamper the diagnostic usefulness of serum RBP4 in patients with liver disease in the real clinical setting. It is well known that the liver is the major source of circulating RBP4 in humans,3 and therefore the hepatic biosynthetic capacity may greatly influence serum RBP4. Although Petta's study excluded patients with Child B and Child C cirrhosis, we could demonstrate in a large cohort of patients with chronic liver diseases of various origins6, 7 that serum RBP4 significantly decreased with the stage of liver cirrhosis (Fig. 1A)).4 Moreover, these differences remained also significant, if patients without cirrhosis were compared to healthy controls or to patients with Child A cirrhosis, which were the subgroups of patients included in the current study by Petta et al. Concordantly, serum RBP4 closely correlated with indicators of liver function in our analysis, such as cholinesterase (r = 0.639), albumin (r = 0.482), or coagulation factors II (r = 0.641) or VII (r = 0.647, all P < 0.001), in patients with chronic liver diseases.4 We corroborated this finding in an animal model of liver fibrosis, as the hepatic RBP4 messenger RNA expression was significantly reduced in rats after surgical biliary obstruction compared to sham-operated controls.4 Interestingly, these findings were confirmed in an independent study by Bahr et al., attributing reduced serum RBP4 in patients with liver cirrhosis to a decreased hepatic RBP4 synthesis.5
We now performed an additional analysis, comparing patients with absent to mild (≤30%) and moderate to severe (>30%) steatosis as determined by liver histology as suggested by Petta et al. Nonetheless, we could not confirm the reported increase in serum RBP4 concentrations in our patients (Fig. 1B). Instead, the degree of liver fibrosis and cirrhosis was the major histological parameter associated with reduced serum RBP4 (Fig. 1C).4 Patients with viral hepatitis in our study did not significantly differ from other etiologies of liver disease (data not shown).
We are well aware that the patient cohort of Petta's work considerably differed from our study, most likely accounting for the observed discrepancies. Whereas Petta et al. included patients without cirrhosis with either HCV infection or NASH, a condition in which elevated serum RBP4 has been described,8 we studied nondiabetic patients with various chronic liver diseases during evaluation for liver transplantation, comprising about 29% patients with viral hepatitis, none with NASH and 52% being at advanced stages (Child B and C) of liver cirrhosis.4 However, we would like to emphasize that the findings of Petta et al. should be perceived with caution, because they may be restricted to a narrow range of specific etiologies of liver diseases (HCV and NASH) and only apply to patients with fully preserved liver function.