Insulin Resistance in Liver Cirrhosis Is Not Associated With Circulating Retinol-Binding Protein 4

Written informed consent was obtained from all participants, and the study protocol was approved by the local ethics committee. We studied 111 Caucasian CLD patients who were evaluated as inpatients for potential liver transplantation (9). Patients with overt diabetes were excluded; none of the patients received insulin or oral antidiabetes/insulin-sensitizing medication or had a history of taking these medications. BMIs for CLD patients were calculated by subtracting the rated volume of ascites and pleural effusions (by ultrasound) from body weight (10).

Eighteen patients (16%) suffered from CLD without having cirrhosis, according to Child-Pugh score and clinical criteria (Table 1); among these patients, the majority was evaluated for liver transplantation due to malignancies or hereditary disorders. The Child-Pugh and the Model for End-Stage Liver Disease (MELD)scores were assessed as independent prognostic predictors in patients with cirrhosis (11). Liver biopsies were performed in 65/111 patients and semiquantitatively evaluated by a blinded and experienced pathologist (12). Ninety-nine age- and sex-matched blood donors (Table 1) with normal aminotransferases, blood counts, fasting glucose, and negative markers for viral hepatitis and HIV served as control subjects.

Serum RBP4 concentrations were measured by enzyme-linked immunosorbent assay (ALPCO Diagnostics) (2), with intra- and interassay coefficients of variation of 5 and 9.8%, respectively. Serum resistin, adiponectin, and C-peptide were measured as described (9,12). Insulin sensitivity was assessed by homeostasis model assessment index (HOMA-S), calculated from fasting glucose and C-peptide (13).

The common biliary duct was double-ligated under halothane anesthesia in male Sprague-Dawley rats (14). Age- and sex-matched sham-operated animals served as controls. Rats were killed after 14 days, and Sirius Red staining was performed on liver tissue sections to determine liver fibrosis (14). All animal experiments were approved by the German authorities.

RBP4 expression was quantified from normal and cirrhotic rat liver by real-time PCR (LightCycler; Roche) using two primer combinations (5′-TGCAGGGTGAGCAGCTTCAG-3′ and 5′-CACTTCCCAGTTGCTCAGAAG-3′ or 5′-TTAGCTCTCATCCAGTCTTC-3′ and 5′-GGAATCCCAAGCCTCAAACG-3′). The average of the measured crossing points (C_T values) was then normalized to glyceraldehyde-3-phosphate dehydrogenase expression (15). The fold increase in mRNA between the experimental groups was calculated by 2^exp(Δ-Δ C_T).

Correlation analyses were done by Spearman's rank correlation test. Comparisons between two groups were conducted with Mann-Whitney U test, multiple comparisons with Kruskal-Wallis ANOVA, and post hoc analysis with Mann-Whitney U tests. Statistical analyses were performed using SPSS (Chicago, IL).

The median serum concentration of RBP4 in healthy control subjects was 35.5 mg/l (range 10.5–85.0), as anticipated from previous literature using the same assay (2). Lean control subjects (median BMI 22.8 kg/m² and RBP4 29.3 mg/l) had significantly lower RBP4 compared with obese control subjects (median BMI 28.4 kg/m² and RBP4 37.5 mg/l, P < 0.001). Patients with CLD showed significantly reduced RBP4 than control subjects (median 11.5 mg/l [range 1.0–117.0], P < 0.001) (Table 1), even if strictly matched to BMI. The decrease in RBP4 was directly related to the stage of liver cirrhosis, as defined by the Child-Pugh score (r = −0.467, P < 0.001) (Fig. 1A). Whereas CLD patients without liver cirrhosis (n = 18, median 30.0 mg/l) did not differ from healthy control subjects, RBP4 significantly decreased between all stages of cirrhosis, with the lowest level in Child C cirrhosis (median 4.3). RBP4 also inversely correlated with the MELD score (r = −0.264, P = 0.005) (Fig. 1B), as an alternative assessment of disease severity. RBP4 was not associated with the patients’ ages or sex (data not shown).

In addition, serum RBP4 directly correlated with the liver's biosynthetic capacity (all P < 0.001), e.g., cholinesterase activity (r = 0.639) (Fig. 1C); serum albumin (r = 0.482); and coagulation factors II (r = 0.641), V (r = 0.633), VII (r = 0.647), and XIII (r = 0.495); and inversely correlated with the prothrombin time (r = −0.546). No correlation was observed between RBP4 and renal function, e.g., serum creatinine or creatinine clearance (data not shown).

As we and others previously described (4,9), insulin resistance is common among CLD patients and clearly linked to the severity of liver cirrhosis. Hyperinsulinemia, reduced insulin sensitivity (HOMA-S), and elevated serum resistin were related to progression of cirrhosis (Table 1). In obese volunteers or diabetic patients with normal liver function, serum RBP4 was described to directly correlate with insulin resistance, impaired glucose tolerance, or BMI (2). In the 18 CLD patients without cirrhosis, we also observed this relationship, as RBP4 correlated with fasting glucose and insulin secretion (C-peptide) and inversely correlated with insulin sensitivity (HOMA-S) (Fig. 2). However, in the 93 patients with Child A-C cirrhosis, no correlation was found for RBP4 with fasting glucose, C-peptide, HOMA index(Fig. 2), BMI, or adipocytokines such as adiponectin or resistin (data not shown).

Decreasing serum RBP4 levels were closely associated with the histological degree of fibrosis or cirrhosis (Fig. 3A) but not with hepatic steatosis (data not shown). Furthermore, clinical complications typically found in patients with advanced cirrhosis were also associated with reduced RBP4 levels: e.g., ascites (median RBP4 5.5 vs. 14.5 mg/l, for patients with or without ascites, respectively, P < 0.001) (Fig. 3B), splenomegaly (9.5 vs. 21 mg/l for patients with or without splenomegaly, respectively, P < 0.001), or esophageal varices (for patients with or without esophageal varices, respectively, 8.0 vs. 17.3 mg/l, P < 0.001) (Fig. 3B). No difference was found in patients with (n = 15) or without (n = 96) hepatocellular carcinoma (data not shown).

As these data collectively indicated that serum RBP4 is linked to hepatic function, we analyzed RBP4 gene expression by real-time PCR from normal and cirrhotic rat liver 14 days after bile-duct ligation (Fig. 3C). RBP4 mRNA expression was ∼3.5-fold higher in normal compared with cirrhotic liver (P = 0.006) (Fig. 3C).

Recent data from mouse models, healthy human volunteers, and patients with obesity, impaired glucose tolerance, or type 2 diabetes described a novel function for RBP4 in the development of insulin resistance (1–3). In our control subjects and the subset of patients without cirrhosis and normal liver function, our observation of correlations between RBP4 and glucose, insulin secretion, or reduced insulin sensitivity is generally in agreement with the recent literature (2,3).

However, in patients with liver cirrhosis and reduced hepatic biosynthetic capacity, no correlation between RBP4 serum levels and insulin resistance, a common finding in advanced disease (4,5,9), exists. In contrast, RBP4 is then closely linked to biomarkers of liver synthesis function and decreases with the progression of disease. This is in line with previous results showing that the liver is the primary source of RBP4 synthesis (1,16,17), despite the reported correlation between adipose RBP4 expression and serum RBP4 (2). Our analysis of an animal model corroborated high gene expression of RBP4 in hepatic tissue and a reduction after induction of experimental cirrhosis.

These results may indicate that RBP4 is, unlike in obesity or type 2 diabetes, not a relevant factor in the pathogenesis of insulin resistance in liver cirrhosis. In contrast to RBP4, several “classical” mediators of insulin resistance were also relevant in patients with liver cirrhosis: hyperinsulinemia or alterations in the adipose-derived adipokines adiponectin and resistin (4,5,9,12). However, whether reduced RBP4, due to reduced hepatic expression by the cirrhotic liver, has any consequences in regulating glucose metabolism in CLD patients is currently unclear. It could potentially render skeletal muscle more susceptible for insulin signals and reduce hepatic gluconeogenesis (1), thereby counteracting factors promoting insulin resistance in cirrhosis. Moreover, vitamin A status, likely reduced in cirrhotic patients (16,17), would influence RBP4's effects on insulin resistance, as RBP4 needs vitamin A to regulate liver enzymes, such as hepatic PEPCK (1). Further investigations are needed to determine the relevance of RBP4 for insulin resistance in liver cirrhosis.

Our findings further imply that RBP4 may not be a clinically useful marker in the presence of concomitant liver dysfunction for indicating overt or the risk for developing insulin resistance. The recent characterization of RBP4 as a reliable biomarker for overt or impending insulin resistance in humans (2,3) is based on the correlation of adipose RBP4 expression and subsequent changes in the muscular and hepatic glucose metabolism (1). However, our study now emphasizes that liver function tremendously impacts serum RBP4. Future studies will need to take liver function into account when examining serum RBP4 levels.

This research project was supported by the START Program of the Faculty of Medicine, RWTH Aachen (to F.T.).