Insulin resistance is associated with chronic hepatitis C and virus infection fibrosis progression
December 2003, Volume 125, Number 6
Jason M. Hui
Background & Aims: Chronic hepatitis C virus infection is associated with an increased prevalence of type 2 diabetes. We hypothesized that virus-induced insulin resistance may be a mechanism for fibrogenesis in chronic hepatitis C virus infection.
Methods: In 260 hepatitis C virus-infected subjects, we examined the relationship between histological findings and anthropometric and biochemical data, including insulin resistance determined by the homeostasis model assessment (HOMA-IR). We also compared fasting serum insulin, C peptide, and HOMA-IR levels between the subset of 121 hepatitis C virus patients with stage 0 or 1 hepatic fibrosis and 137 healthy volunteers matched by sex, body mass index, and waist-hip ratio.
Results: Hepatitis C virus-infected subjects with stage 0 or 1 hepatic fibrosis had higher levels of insulin, C peptide, and HOMA-IR (all P<=0.01) compared with matched healthy controls. In the 250 hepatitis C virus patients (fibrosis stage 0 to 4), viral genotype and portal, but not lobular, inflammation were univariate predictors of HOMA-IR. By multiple linear regression analysis, independent predictors of HOMA-IR included body mass index (P < 0.001), previous failed antiviral treatment (P < 0.001), portal inflammatory grade (P < 0.001), and genotype 3 status (P = 0.01). Genotype 3 had significantly lower HOMA-IR than other genotypes (which were comparable when adjusted for effects of the remaining independent predictors). HOMA-IR was an independent predictor for the degree of fibrosis (P < 0.001) and the rate of fibrosis progression (P = 0.03).
Conclusions: Hepatitis C virus may induce insulin resistance irrespective of the severity of liver disease, and this effect seems to be genotype specific. Further, our findings support the hypothesis that insulin resistance may contribute to fibrotic progression in chronic hepatitis C virus infection.
Recent evidence suggests that chronic hepatitis C virus (HCV) infection is associated with an increased risk for the development of type 2 diabetes. Thus, type 2 diabetes is more prevalent among patients with chronic HCV compared with those with other liver diseases and the general population, irrespective of whether cirrhosis is present. Likewise, HCV seropositivity among patients with type 2 diabetes mellitus is higher than in the general population, a finding that has been validated for different ethnic groups. Finally, even if liver function is restored by transplantation, post-liver transplantation diabetes mellitus occurs more frequently among patients who undergo transplantation for HCV than for other conditions.
Insulin resistance (IR) plays a primary role in the development of type 2 diabetes mellitus. This is supported by prospective longitudinal studies showing that IR is the best predictor for the development of diabetes, preceding the onset of diabetes by 10 to 20 years, and by cross-sectional studies showing that IR is a consistent finding in patients with type 2 diabetes. In view of the strong association between HCV infection and the development of diabetes mellitus, it is important to determine whether HCV infection can predispose to the development of IR before diabetes occurs. Such a potential link is particularly cogent in light of recent data that indicate that diabetes may be associated with increased fibrosis progression in patients with chronic HCV infection. In this study, we tested the hypothesis that HCV infection itself may promote IR, by comparing the degree of IR (determined by fasting glucose, insulin, and C peptide levels and the homeostasis model [HOMA-IR]) between HCV-infected individuals and healthy volunteers. We then examined whether histological markers of HCV activity (portal and lobular inflammation) were associated with HOMA-IR and whether there were genotype-specific alterations in the extent of HOMA-IR. Finally, we assessed whether such virus-induced IR may be a mechanism for fibrogenesis in chronic HCV infection, by determining the relationship between the magnitude of HOMA-IR and the severity of hepatic fibrosis and the rate of fibrosis progression.
This study comprised 260 consecutive patients with chronic HCV who underwent liver biopsy at Westmead Hospital between May 1999 and August 2002. Some of these cases (n = 117) have been the subject of a previous report. All subjects had antibodies against HCV (Monolisa anti-HCV; Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France) and detectable HCV RNA by polymerase chain reaction (Amplicor HCV; Roche Diagnostics, Branchburg, NJ). HCV genotyping was performed with a second-generation reverse hybridization line probe assay (Inno-Lipa HCV II; Innogenetics, Zwijndrecht, Belgium). Thirty-three subjects (13%) had previous antiviral therapy (interferon or pegylated interferon monotherapy [n = 30] or combination therapy with interferon and ribavirin [n = 3]) and either were nonresponders (n = 20) or had relapsed after treatment (n = 13). Liver biopsies were performed at least 6 months after the completion of antiviral therapy (median, 5 years; range, 0.5--10 years). No patient had clinical evidence of hepatic decompensation (hepatic encephalopathy, ascites, variceal bleeding, or serum bilirubin level greater than 2-fold the upper limit of normal).
Clinical and laboratory assessment
The following data were collected at the time of liver biopsy: age, sex, ethnicity, average current daily alcohol intake (g/day) in the past 6 months, past alcohol intake (g/day) before the last 6 months, weight, height, and waist-hip ratio (WHR; waist circumference at umbilicus/hip circumference at the maximal circumference over the buttocks). Body mass index (BMI) was calculated as weight in kilograms/height in square meters. Past exposure to HBV was determined by the presence of HBV core antibody.
The estimated duration of infection was defined as the time elapsed from the presumed date of infection to the date of liver biopsy. The former was estimated as follows: the date of transfusion of blood products (before 1990), the first year of intravenous drug use, or the date of a single specific and convincing parenteral exposure (e.g., needlestick injury). We defined fibrosis progression per year as the ratio of the fibrosis stage by the Scheuer score to the estimated duration of infection in years.
After an overnight fast of 12 hours, venous blood was drawn to determine the serum levels of albumin, bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), ALT/AST ratio, -glutamyltransferase (GGT), ferritin, cholesterol, insulin and C peptide, plasma glucose concentration, platelet count, and international normalized ratio. Serum insulin was determined by radioimmunoassay (Phadaseph Insulin RIA; Pharmacia and Upjohn Diagnostics AB, Uppsala, Sweden). Serum C peptide was estimated by a competitive immunoassay (IMMULITE; Diagnostic Products, Los Angeles, CA). All other biochemical tests were determined by automated procedures in the clinical pathology laboratories of Westmead Hospital. IR was determined by the homeostasis model assessment (HOMA) method by using the following equation: Insulin resistance(HOMA-IR)=Fasting insulin(µU/mL)xFasting glucose(mmol/L)/22.5 IR calculated by this method has been validated against insulin sensitivity measured directly with the euglycemic/hyperinsulinemic clamp technique in both diabetic and nondiabetic subjects. To take into account the effect of advanced hepatic fibrosis on increasing serum insulin levels that is partly due to impaired insulin clearance, we determined insulin secretion by using the serum C peptide-insulin ratio. C peptide and insulin are secreted in equimolar amounts, and serum C peptide is not significantly cleared by the liver; hence, the C peptide-insulin ratio allows hyperinsulinemia due to impaired insulin degradation (low ratio) to be distinguished from insulin hypersecretion (normal ratio).
The following conditions were excluded, as described previously: concurrent active hepatitis B virus (HBV; positive for hepatitis B surface antigen) or human immunodeficiency virus infection, autoimmune hepatitis, primary biliary cirrhosis (PBC), sclerosing cholangitis, hemochromatosis, 1-antitrypsin deficiency, and Wilson’s disease. Patients with an established diagnosis of diabetes mellitus were excluded from this study. The study protocol was approved by the Human Ethics Committee of the Western Sydney Area Health Service, and written, informed consent was obtained.
The degree of necroinflammatory activity and fibrosis were scored semiquantitatively as described by Scheuer by an experienced hepatopathologist ( J. G. K.) blinded to the clinical data. Portal or periportal and lobular inflammatory activities were both scored from 0 to 4. Fibrosis was scored as follows: F0, no fibrosis; F1, enlarged fibrotic portal tracts; F2, periportal or portal-portal septa, but intact architecture; F3, architectural distortion but no obvious cirrhosis; and F4, probable or definite cirrhosis. Steatosis was assessed as the percentage of hepatocytes containing macrovesicular fat droplets. It was graded as 0 (no steatosis), 1 (<33% of hepatocytes affected), 2 (33%--66% of hepatocytes affected), or 3 (>66% of hepatocytes affected).
Case controls: healthy volunteers
One hundred thirty-seven apparently healthy volunteers with normal liver function tests and no known history of diabetes mellitus were enrolled. They were matched by sex, BMI, and WHR with the 121 HCV-infected subjects with minimal (stage 1) or no (stage 0) hepatic fibrosis. Markers of IR (fasting glucose, insulin, HOMA-IR, and C peptide) were compared between these 2 groups of matched subjects.
Case controls: primary biliary cirrhosis
Twenty-four subjects with PBC and no known history of diabetes mellitus were identified from the Westmead Hospital database. The liver biopsy samples of these subjects were reviewed (by J. G. K). To allow comparison with the HCV cases, the fibrosis stage of these biopsy samples was scored by Scheuer’s method; all had fibrosis stages 1 to 3, and none had cirrhosis. The markers of IR (fasting glucose, insulin, HOMA-IR, and C peptide) were compared among the PBC subjects, the HCV patients with stage 0 fibrosis, and the 137 healthy volunteers.
Patient characteristics and liver biopsy findings
The mean age was 41 ± 9 years (range, 16--72 years); 175 (67%) were male, and the mean BMI was 26.9 ± 5.2 kg/m2 (range, 17.2--47.3 kg/m2). HCV genotype was available in 250 patients. Of the remaining 10 patients, 5 were not typeable by the line probe assay, and 5 were not tested. These 10 cases were missing at random and were excluded from subsequent analyses involving genotype. Because liver biopsy specimens were required to show at least 4 portal tracts for reliable scoring, complete histological analysis of necroinflammatory grade and fibrosis staging was available in only 258 cases; however, steatosis grade could be assessed in all 260 cases. Fibrosis was absent in 39 (15%) patients, was stage 1 in 82 (32%) patients, was stage 2 in 85 (33%) patients, and was stage 3 in 30 (12%) patients, whereas cirrhosis (stage 4 fibrosis) was present in 22 (9%) patients.
Reliable data on the date of infection were available for 117 (45%) patients. The mean age of infection was 22 ± 7 years; the mean estimated duration of infection was 19 ± 8 years, and the mean rate of fibrosis progression was 0.13 ± 0.21 stages per year.
Insulin resistance of hepatis C virus cases compared with matched healthy volunteers
To assess the influence of HCV infection on IR independent of any effect of hepatic fibrosis, 121 HCV subjects with minimal (stage 1) or no (stage 0) fibrosis were compared with 137 healthy volunteers matched by sex, BMI, and WHR. Ideally, a comparison population would also be age matched, because the frequency of IR increases with age. However, even though the HCV subjects were younger than the healthy volunteers, they had significantly higher levels of all markers of IR, including fasting glucose, insulin, C peptide, and HOMA-IR.
Viral factors associated with the degree of insulin resistance
To determine the possible virus-related factors involved in the pathogenesis of IR, we assessed whether HOMA-IR was associated with the viral genotype or the severity of portal or periportal and lobular inflammation after controlling for other demographic and biochemical variables. By univariate analysis, the factors associated with HOMA-IR were portal or periportal inflammatory grade, viral genotype, age, BMI, WHR, GGT, albumin, bilirubin, C peptide, and previous treatment (either nonresponders or relapse after treatment—see Methods. Lobular inflammatory grade was not associated with HOMA-IR (P = 0.6).
The final model for the independent predictors of HOMA-IR by multiple linear regression analysis included portal or periportal inflammatory grade, genotype 3 status (yes or no), BMI, and previous treatment. These variables explained 30% of the variability in the degree of HOMA-IR.
At each stage of fibrosis, the genotype 3 patients had higher (unadjusted) mean HOMA-IR than the non-genotype 3 patients. After adjusting for the effect of other independent predictors in the model, the various non-genotype 3 groups had comparable HOMA-IR levels, and the adjusted estimated difference in HOMA-IR between the genotype 3 and non-genotype 3 subjects was --0.58 (95% confidence interval [CI], --0.13 to --1.02; P = 0.01). There was no significant interaction between the effects of genotype 3 status and fibrosis stage, either unadjusted (P = 0.5) or adjusted (P = 0.4) for the other independent variables.
Because cirrhosis is known to cause IR, the multivariate analysis was repeated for the subset of 236 noncirrhotic subjects (fibrosis stages 0 to 3).
The independent predictors for HOMA-IR were the same as for all subjects: genotype 3 status (P = 0.005), portal inflammatory grade (P = 0.007), BMI (P < 0.001), and previous treatment (P < 0.001) remained in the model.
Factors associated with the severity of hepatic fibrosis and the rate of fibrosis progression
We next assessed whether the extent of IR was associated with the severity of hepatic fibrosis. By univariate analysis, factors associated with the stage of fibrosis were age, male sex, past alcohol intake, WHR, ALT, AST, GGT, albumin, bilirubin, platelet count (negative association), international normalized ratio, glucose, insulin, C peptide, HOMA-IR, ferritin, and cholesterol (negative association), portal or periportal inflammatory grade, lobular inflammatory grade, steatosis, and viral genotype. By multiple ordinal regression analysis, independent predictors for the degree of fibrosis were portal or periportal inflammatory grade, past alcohol intake, HOMA-IR, age, and ALT; platelet count and cholesterol were significant negative predictors. Together, these factors accounted for 52% of the variability in hepatic fibrosis stage. Multiple linear regression analysis with stepwise variable selection confirmed the fitted model.
The probability of moderate to severe fibrosis (stage 2 to 4) was 45% for those with a HOMA-IR of 1 compared with 68% for those with a HOMA-IR of 5 (evaluated at the mean value of other independent predictors: portal inflammatory grade, 2.0; past alcohol intake, 10--40 g/day; age, 41.1 years; ALT, 115 U/L; platelet count, 234; and cholesterol, 4.4 mmol/L). Although steatosis was associated with fibrosis by univariate analysis, it was not an independent predictor of fibrotic stage after adjusting for other factors.
Because genotype 3 subjects had significantly lower HOMA-IR, subgroup analyses were performed to determine whether HOMA-IR was an independent predictor for fibrosis in both genotype 3 and non-genotype 3 subjects. By multiple ordinal regression analysis, HOMA-IR remained independently associated with the fibrosis stage for both the genotype 3 (OR, 1.4; 95% CI, 1.0--2.0; P = 0.03) and non-genotype 3 subgroups (OR, 1.2; 95% CI, 1.1--1.4; P = 0.007).
Because cirrhosis is known to cause IR, the multivariate analysis for predictors of fibrosis stage was repeated on the 236 noncirrhotic subjects (fibrosis stages 0 to 3). The independent predictors for fibrosis were unchanged, and HOMA-IR remained in the model (OR, 1.3; 95% CI, 1.1--1.5; P < 0.001). Furthermore, there was no significant association between the C peptide-insulin ratio and the fibrosis stage for the noncirrhotic patients (r = --0.1; P = 0.1).
To determine whether HOMA-IR was a cause or a consequence of hepatic fibrosis, we analyzed the subgroup of 117 patients with a known duration of infection. By multiple linear regression analysis, HOMA-IR was independently associated with an increased rate of fibrosis progression.
Factors associated with the extent of hepatic steatosis
To examine the hypothesis that IR was associated with fibrosis through an effect on hepatic steatosis, we performed a multiple ordinal regression analysis to determine independent predictors for steatosis grade. Independent predictors for hepatic steatosis grade were genotype 3 status (P < 0.001), BMI (P = 0.002), portal inflammation (P = 0.03), and cholesterol (negative association; P = 0.004). Genotype 3 status remained a significant factor in the model (P < 0.001) even when cholesterol was not entered into the model. Although HOMA-IR was not an independent predictor for steatosis grade, it should be noted that BMI was associated with HOMA-IR (r = 0.5; P < 0.001).
Insulin resistance of primary biliary cirrhosis cases compared with hepatitis C virus cases and healthy volunteers
We next considered the possibility that hepatic fibrosis from any etiology, even in the absence of cirrhosis, may cause IR. We therefore examined markers of IR in 23 noncirrhotic PBC subjects in whom fibrosis was staged according to Scheuer (11 with stage 1 fibrosis, 10 with stage 2 fibrosis, and 2 with stage 3 fibrosis) and compared the results with those of the 39 HCV subjects with no hepatic fibrosis (stage 0) and the 137 healthy volunteers. After controlling for the effects of BMI and sex, HCV subjects with stage 0 fibrosis were found to have higher levels of serum insulin (P = 0.01), C peptide (P < 0.01), and HOMA-IR (P = 0.01) than patients with PBC or healthy controls. However, markers of IR were not significantly different between the PBC subjects and healthy volunteers.
DISCUSSION by authors
This study, in subjects without a history of diabetes mellitus, shows that HCV infection increases IR compared with healthy volunteers. There are genotype-specific effects on IR, and the extent of portal inflammation in chronic HCV is associated with the degree of IR. Furthermore, increased HOMA-IR values are associated with a higher rate of fibrosis progression and more advanced stages of hepatic fibrosis. These data support the concept that viral-induced IR may facilitate fibrogenesis in chronic HCV infection.
Use of the homeostasis model assessment model
The HOMA model used in this study has been validated and widely used for determining the degree of IR in epidemiological studies. HOMA-IR accounts for approximately 65% of the variability in insulin sensitivity assessed by the glucose clamp technique. It seems to be as good a predictor of clamp-determined insulin sensitivity as the short insulin tolerance test or the intravenous glucose tolerance test analyzed with the minimal model. HOMA-IR strongly predicts the development of type 2 diabetes, independent of obesity, body fat distribution, and glucose tolerance status.
Evidence that hepatitis C virus infection is specifically associated with insulin resistance
In this study, HCV-infected subjects in whom there was no or minimal hepatic fibrosis were closely matched to healthy volunteers by sex and anthropometric predictors of IR, namely, BMI and WHR. Although the HCV-infected subjects were younger than the volunteers, fasting serum insulin, C peptide, and HOMA-IR were greater in those with HCV infection. Earlier studies, which found increased fasting insulin levels and reduced insulin sensitivity in HCV-infected subjects, included patients with moderate to severe hepatic fibrosis. The present data extend these findings to HCV-infected subjects with minimal or no fibrosis.
In this study, the grade of portal inflammation, a hallmark of chronic HCV infection that correlates with fibrotic progression, was associated with HOMA-IR in nondiabetic subjects. This suggests that the virus itself or the host inflammatory response to HCV infection contributes to the development of IR and increases the long-term risk for the development of type 2 diabetes. We also considered the converse possibility, that is, IR plays a pathogenic role in hepatic necroinflammation, as is the case for nonalcoholic steatohepatitis (NASH). However, we believe this to be unlikely because there was no association among the grade of lobular inflammation, the predominate site of inflammation in NASH, and HOMA-IR. Further, the findings of 2 small studies that insulin sensitivity and glucose tolerance improve with antiviral therapy are consistent with our proposal that IR is mediated by viral infection or the inflammatory response to HCV, rather than the converse.
It was of interest that patients with previously failed antiviral treatment had significantly higher IR, independent of BMI, than those who had not received previous antiviral therapy. It has been suggested that the impaired response to antiviral therapy in African Americans may relate to their high rate of central obesity and IR. Prospective studies to assess whether IR is an independent predictor for a reduced response to antiviral therapy are required. This is important because implementation of lifestyle changes can improve insulin sensitivity.
The mechanisms for IR in chronic HCV infection remain unclear. In a recent report, hyperinsulinemic clamp assessment in 5 HCV-infected subjects showed reduced glucose disposal consistent with peripheral IR. Activation of the tumor necrosis factor (TNF) system occurs in chronic HCV infection and correlates with disease activity, and higher levels of TNF expression may be predictive of a failed response to interferon therapy. TNF also plays an important role in the pathogenesis of IR and may provide the pathogenic link between chronic HCV infection and the IR we observed.
The effect of hepatitis C virus infection on insulin resistance varies between genotypes
An important novel finding of this work is that the effect of HCV infection on IR depends on viral genotype. Thus, for each stage of hepatic fibrosis, genotype 3 subjects have lower IR compared with other genotypes, even after adjusting for the effect of BMI and other confounders. This would confer a lower risk for developing type 2 diabetes compared with other HCV genotypes. Indeed, the increased prevalence of diabetes in HCV has been shown to be predominately among genotype 1-- and 2-infected subjects. It is of interest that, despite lower IR levels, subjects with HCV genotype 3 have more extensive hepatic steatosis. The implication that steatosis in genotype 3 is mediated predominately by viral factors and not IR is consistent with earlier reports.
Insulin resistance is associated with more rapid fibrosis progression in chronic hepatitis C virus infection
IR and diabetes mellitus are independent predictors of the severity of hepatic fibrosis in nonalcoholic fatty liver disease; diabetes mellitus is also associated with increased fibrosis in chronic HCV. The most important finding of this study (which excluded those with type 2 diabetes mellitus) is that increased HOMA-IR is a predictor of the stage of fibrosis and the rate of fibrosis progression, independent of other known factors including age, male sex, past alcohol intake, platelet count, and portal inflammation. A recent report found that weight loss can improve hepatic fibrosis in HCV. This is likely to be a result of the improvement in IR associated with weight loss, which is consistent with the proposal that IR is a clinically important determinant of the rate of fibrosis progression in HCV.
An alternative explanation for the observed association between HOMA-IR and fibrosis stage is that hepatic fibrosis led to impaired insulin degradation. Several findings help to exclude that possibility. First, HOMA-IR remained an independent predictor of fibrosis stage even after the exclusion of subjects with cirrhosis, which is known to cause IR and impaired insulin clearance. Second, the observations that C peptide was a univariate predictor of fibrosis stage and that the serum C peptide-insulin ratio was similar in different fibrosis stages indicate that insulin hypersecretion, and not impaired degradation, accounts for the increase in HOMA-IR in HCV infection. Third, markers of IR in PBC, even in the presence of moderate portal fibrosis, were lower than those in nonfibrotic HCV and similar to those in healthy volunteers (after controlling for the effects of BMI). However, we acknowledge the possibility that the factors responsible for IR may also be responsible for causing progressive fibrosis and that the association between HOMA-IR and fibrosis stage does not prove a casual relationship.
In earlier studies, steatosis in chronic HCV infection has been associated with increased fibrosis stage. IR was not determined in any of these reports, but is known to be an important pathogenic factor for steatosis in NASH. Although IR was not an independent predictor of steatosis in our cohort, this may be due to the confounding effect of BMI, a predictor of steatosis that is closely associated with IR.
Plausible mechanisms exist to explain the role of IR in hepatic fibrogenesis. Hyperinsulinemia can directly stimulate hepatic stellate cells to proliferate and to secrete extracellular matrix. Further, high glucose levels and hyperinsulinemia cause up-regulation of connective growth factor, a cytokine involved in the pathogenesis of fibrosing liver diseases.
In conclusion, this study provides the first direct evidence for a genotype-specific association between chronic HCV infection and IR. Our data support the hypothesis that IR may increase the rate of fibrosis progression. Strategies to improve insulin sensitivity should be explored because they may complement antiviral therapy in the management of chronic HCV infection, particularly in mitigating against fibrotic progression in nonresponders to interferon-based antiviral therapies.