Review article: an unexpected virus–host interaction – the hepatitis C virus-diabetes link
Prof. V. Ratziu, Service d'hépatogastroentérologie, Hôpital Pitié Salpêtrière and Université Pierre et Marie Curie, 47-83 Bd de l'Hôpital, Paris 75013, France.
There is now consistent epidemiological evidence for an association between chronic hepatitis C and diabetes. Important, although so far limited longitudinal data, have documented an increased risk for diabetes in patients infected by hepatitis C virus (HCV) especially in those with metabolic risk factors such as a high BMI and older age. HCV encoded proteins might alter insulin signalling thus explaining impaired insulin sensitivity and the occurrence of glycaemic dysregulation even before the cirrhotic stage. The consequences of the association between diabetes and HCV infection are an increased liver fibrosis stage and faster fibrosis progression rate. This article reviews recent human and experimental data on the HCV-diabetes association.
Both hepatitis C and diabetes are common diseases with an estimated 160 million individuals infected by hepatitis C virus (HCV) worldwide and 120 million suffering from diabetes. Besides their considerable prevalence these two diseases seem to share little in common and yet recent data have suggested that their concurrent presence in some individuals may be higher than would be expected by chance alone. Certainly, advanced cirrhosis induces dysregulation of glycaemic control which may result in overt diabetes.1 However, this does not account for the bulk of data from numerous studies from liver clinics documenting a higher prevalence of diabetes among HCV patients than among controls. Other studies from patients followed for endocrine or haematological disorders confirmed the epidemiological association between chronic hepatitis C and type 2 diabetes.
This article will summarize some recent key data on the epidemiological association between these two conditions, their mechanisms and their clinical relevance. Excellent recent reviews are available for a more comprehensive report.2
Does the epidemiological association occur by chance or by design?
Two important studies performed in the general population have documented an increased prevalence of diabetes among patients infected with HCV.3, 4 The first is a multicentric cross-sectional survey in the United States involving 9822 individuals in whom the relative risk of being diagnosed with diabetes increased 3.8 times in the presence of HCV.3 Interestingly, although the association between HCV infection and type 2 diabetes was independent of age, sex, race, body mass index (BMI) and family history of diabetes, it was significant only in subjects older than 40. This suggests that although HCV infection might induce insulin resistance, the occurrence of overt diabetes necessitates additional factors such as the physiological resistance to insulin induced by ageing.
Because of its cross-sectional design the study by Mehta et al.4 cannot determine if there is a causal relationship behind the mere association between HCV infection and diabetes. Therefore, the only longitudinal study available thus far provides critical information in this regard. The same author studied a community-based cohort of 10 275 individuals who were free of diabetes at entry and were followed over a 9-year period. The incidence of diabetes was higher in individuals with HCV infection at entry, but here again, the excessive risk was significant only in older patients with a higher BMI. In this group of individuals with a high risk of diabetes development, those with HCV infection had 12 times more chance of developing diabetes during follow-up. Unfortunately, the prevalence of HCV infection was surprisingly low in this cohort (0.8% vs. 1.8% in the USA population) and thus the study could be underpowered for detecting the impact of HCV infection in patients at low risk for diabetes, namely in younger lean individuals. Nevertheless, the study provides valuable information showing that as HCV infection precedes the development of diabetes, a causal relationship is possible, at least in high-risk individuals.
By selecting individuals in the general population most of whom were presumed to be in good health, these community-based studies minimize the confounding contribution of liver disease per se in the development of diabetes. However, the possibility still exists that advanced liver disease induced by HCV (i.e. cirrhosis) accounts, non-specifically, for the association between HCV and diabetes. Several studies are now available showing that even HCV patients with chronic hepatitis but not cirrhosis have an increased prevalence of diabetes.5–7 Recently, Kawaguchi et al.7 showed higher serum insulin levels and higher homeostatis model assessment (HOMA) values in 158 Japanese patients with non-cirrhotic chronic hepatitis C than in 179 controls with diverse other liver diseases. This suggests that even before the cirrhotic stage, HCV infection induces insulin resistance that can predispose to the development of diabetes. Taking the matter one step further, Lecube et al.6 demonstrated a fourfold higher risk of diabetes and/or impaired fasting glucose in 498 non-cirrhotic HCV patients than in 144 non-cirrhotic patients with non-HCV-related chronic hepatitis. Moreover, in HCV patients, an oral glucose tolerance test was diagnostic of diabetes in 18% and of impaired glucose tolerance in 30%. As this was much higher than in the non-HCV population, the authors went recommended universal screening of glucose metabolic disorders in HCV patients. However, it is important to keep in mind that no study to date has matched HCV and control patients on the degree of liver fibrosis and therefore a higher proportion of bridging fibrosis in the HCV group than in the non-HCV group might account for these differences (see below).
Are patients with HCV infection more insulin-resistant than those with other liver diseases?
If the epidemiological association holds true and if HCV contributes to the occurrence of diabetes even in patients without cirrhosis, then HCV-infected patients should display higher levels of insulin resistance. Several studies are now available comparing surrogate markers of insulin resistance in HCV-infected and non-HCV patients.7–9 Most, but not all,8 have demonstrated higher fasting insulin serum levels and higher HOMA indices in HCV than in non-HCV patients. However, this type of comparison is meaningful only if cases and controls are carefully matched, not only on anthropometric risk factors for diabetes (age, sex, BMI) but also for the degree of liver injury, mainly the fibrosis stage. Indeed, Petit et al. made the important observation in HCV patients that increasing levels of fibrosis are associated with increasing insulin resistance as evaluated by the HOMA model.10Subsequently this observation has been largely confirmed.7–9 One study in particular has shown that HCV patients have higher levels of insulin resistance and of hyperinsulinaemia than controls with primary biliary cirrhosis, even after adjustment for age, sex and BMI and when considering only patients with no or minimal fibrosis.9
What, then, causes hyperinsulinaemia in HCV-infected individuals? A first explanation could be that the hepatic insulin extraction is impaired either because of advanced liver disease or because of coexisting steatosis,11–13 and therefore higher systemic levels of insulin are simply the result of reduced hepatic clearance. This, however, has been ruled out by the demonstration that hepatic insulin extraction, as assessed by the C-peptide/insulin ratio, was similar between HCV and non-HCV patients.14 An alternative explanation could be that liver injury, irrespective of its cause, induces insulin resistance, for instance through cytokines such as tumour necrosis factor (TNF)-α and/or interleukin (IL)-6 which are induced during liver inflammation and/or fibrosis. Hickman et al. argue against this possibility by demonstrating that while there is a correlation between insulin levels and fibrosis stages in overweight HCV patients, no such correlation exists in lean individuals.8 Therefore it appears that, at least in HCV patients, hyperinsulinaemia is not a non-specific effect of progressing liver injury. A third possibility would be that HCV induces insulin resistance by interfering with insulin signalling. Recently, several human and experimental studies have given credit to this theory.
What are the mechanisms of altered insulin signalling by HCV?
Insulin signalling is initiated upon binding to a specific receptor, the insulin receptor, which is expressed in target cells and has a tyrosine kinase activity. This results in the phosphorylation of tyrosine residues of insulin receptor substrate (IRS) proteins, especially IRS-1 and IRS-2, which play a crucial role in the regulation of the multiple growth and metabolic effects of insulin. Upon tyrosine phosphorylation, IRS starts recruiting additional adapter and signalling molecules, resulting in the activation of second messenger cascades.15, 16 One such molecule is phosphatidylinositol 3-kinase (PI3-kinase) and its downstream effector, Akt which control many of the metabolic effects of insulin.17
Aytug et al. first reported impaired IRS-1/PI3-kinase signalling in patients with HCV infection.18 They demonstrated reduced tyrosine phosphorylation of IRS-1 in liver tissue of patients with HCV infection compared to controls with non-viral liver disease. This was accompanied by a significant decrease in insulin-stimulated PI3-kinase activity and a severely blunted Akt phosphorylation. Further confirmation of this important study is eagerly awaited.
Two important sets of experimental data confirm these human findings and demonstrate that the core protein of HCV are responsible for these effects. Kawaguchi et al. performed elegant in vitro studies demonstrating that the core protein of HCV reduces IRS-1 and IRS-2 expression with resulting inhibition of insulin signalling through the PI3-kinase pathway and reduced Akt-dependent metabolic effects.7 The reduction in IRS-1 and -2 expression was the result of the induction by the core protein of SOCS3, which in turn promotes proteosomal degradation of IRS-1 and -2 through ubiquitination.7 Working with transgenic mice for the HCV core protein, Shintani et al. demonstrated in vivo that overexpression of the core protein induces hepatic insulin resistance.19 Indeed, HCV core transgenic mice compared with wild-type animals displayed high serum insulin levels and impaired inhibition of hepatic glucose output in response to insulin. Remarkably, these changes were documented at a young age, before the occurrence of steatosis which by itself could cause insulin resistance.20, 21 This hepatic insulin resistance precipitated the occurrence of overt diabetes upon feeding the mice a high fat diet. Expression of the core protein transgene resulted in altered tyrosine phosphorylation of IRS-1 (but not IRS-2) and increased hepatic expression of TNF-α, two possible reasons why these mice developed insulin resistance. In vivo antagonism of TNF-α by systemic administration of anti-TNF-α antibodies restored insulin sensitivity and tyrosine phosphorylation of IRS-1.19 Data in humans are thus far limited but preliminary results suggest that the level of circulating HCV core antigen correlates with hyperinsulinaemia in HCV patients.7
However, enticing all these models might be, they do not entirely accommodate the available data on differential tissue distribution and function of individual IRS. Indeed, the liver does not appear to be the main target of IRS-1 because glycogen synthesis and hepatic glucose output are mainly regulated by IRS-2.22 Moreover, IRS-1 deficiency through genetic inactivation is very effectively compensated by IRS-2, at least in the liver.23 Thus, additional data are needed before drawing firm conclusions on the mechanisms and intermediaries of HCV-induced insulin resistance.
Is the association between diabetes and HCV infection clinically significant?
Several studies have now documented the deleterious impact of insulin resistance on liver injury. Insulin resistance as measured by the HOMA model has a significant albeit rather modest contribution to fibrosis stage and fibrosis progression9, 24 which is independent of age and genotype.9 Interestingly this holds true even before the cirrhotic stage.9 As stated above, high serum insulin levels are also associated with increasing fibrosis stages,24 at least in overweight patients.8 High serum glucose has been recognized as an independent predictor of liver fibrosis even after adjustment for age, sex and alcohol consumption.25 Its profibrogenic effect is stronger and independent from that of the BMI and again, is noticeable even before the cirrhotic stage.25 Finally, several studies have also documented that overt diabetes is independently associated with more severe fibrosis in patients with chronic hepatitis C.26, 27
The association of insulin resistance to fibrosis raises nevertheless other interesting questions. One of them is what are the mechanisms explaining increased fibrogenesis in insulin-resistant states?28 While there are no definitive answers, certain hypotheses are worth pursuing in future studies. Insulin resistance could contribute to a chronic inflammatory state at the tissue level with resultant expression of proinflammatory cytokines such as TNF-α.29 Alternatively, hyperinsulinaemia can induce proliferation of hepatic stellate cells30 and high levels of glucose induce the expression of connective tissue growth factor,31 a strongly profibrogenic cytokine. Another relevant mechanism is related to the presence of advanced glycated end products (AGE). These are the result of glycosylation of a large number of serum proteins induced by long-standing hyperglycaemia. Receptors for AGE have been demonstrated on the cell membrane of hepatic stellate cells and ligand-induced activation of these receptors triggers the myofibroblastic activation of quiescent stellate cells.32
Another important question is that regarding the relative contribution of steatosis per se vs. that of insulin resistance in the genesis and progression of liver fibrosis. Obviously these two conditions are linked and recognizing their specific contributions is difficult. Almost all studies have documented an association between steatosis and fibrosis by univariate analysis28 and some have suggested a relation between steatosis worsening and fibrosis progression.33 Experimental data have demonstrated the activation of profibrogenic pathways in the presence of steatosis.34–36 However, the role of steatosis may not be direct and may be confounded by the underlying insulin resistance.37 Indeed, after taking into account all measurable risk factors of insulin resistance the independent association between steatosis and fibrosis is no longer significant in some studies.9, 25, 38
This is an era of fascinating discoveries of the complex interplay between the virus and host metabolic pathways. The association between HCV infection and type 2 diabetes is real and appears to be causally linked, at least in predisposed individuals (older and overweight). The virus itself, and not the liver disease may be the culprit by interfering with insulin signalling. As a consequence, patients with HCV infection appear to be more insulin-resistant than those with other types of liver disease, even after adjusting for the stage of fibrosis. Clinically, long-standing insulin resistance, hyperglycaemia and diabetes may worsen liver fibrosis. Whether it also reduces the efficacy of anti-viral treatment remains to be studied but this may be the case as overweight and steatosis are associated with a lower response.39–41 Preliminary data suggest that correction of insulin resistance might help achieve higher response rates with anti-viral treatment.42, 43