A total number of 193 patients who met the criteria were eligible to enter the study. First group included 41 cirrhotic patients. Mean age was 52.1 ± 11.4 years ranging between 26–74 years. Diagnosis of cirrhosis was confirmed by histology in 17 (41.5%) patients. The occurrence of signs or biochemical evidences of liver decompensation, ultrasound features of portal hypertension and/or esophageal varices in gastroscopy were used for the clinical diagnosis in the remaining cases. No cirrhotic patients had evidence of hepatocellular carcinoma, screened by serum alpha-fetoprotein level test and abdominal ultrasonography. Severity of liver cirrhosis was graded according to Child-Pugh's classification. Second group included 102 patients with chronic hepatitis B, or C. Mean age was 41.2 ± 10.9 years ranging between 22–67 years. All patients had adequate documentation of elevated serum aminotransferases for more than 6 months, and viral hepatitis molecular assays by qualitative RT-PCR. In 91 (89.2%) patients liver biopsy reports were accessible. The histological staging and grading in liver biopsy was scored according to Knodell scoring. Third group included 42 inactive hepatitis B virus (HBV) carriers who were hepatitis B s antigen (HBsAg) positive, quantitative HBV DNA level less than 10⁵ copies/ml (ROCHE Cobas Amplicore) and normal liver enzymes. The mean age in this group was 42.2 ± 12.0 years ranging between 23–84 years. Eight patients with nonalcoholic steatohepatitis (NASH) were initially enrolled but were not entered in analysis because of the low case number. Therefore analysis was finally performed for 185 patients.

Informed consent in writing was obtained from each patient involved and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as approved by institutional review board. Initial data were collected by chart review and interview. Then, two venous blood samples were taken: First sample in the morning after a 12-hour overnight fast to measure fasting plasma glucose (FPG) level, triglycerides (TG) level, and cholesterol (Chol) level; and a second sample following 2 hours after eating 75 gr glucose to measure two-hour post-loaded glucose (2-hr PG). FPG and 2-hr PG measurement were repeated on two other samples in second session.

Patients receiving therapy with insulin or oral hypoglycemic medications were considered as diabetics. In other patients, FPG ≥ 126 mg/dl or 2-hr PG ≥ 200 mg/dl on more than one occasion was used as diagnostic for DM in agreement with the latest American Diabetes Association (ADA) criteria [13]. FPG between 110 mg/dl and 126 mg/dl was considered as impaired fasting glucose (IFG), and 2-hr PG between 140 mg/dl and 200 mg/dl was considered as IGT according to ADA definitions [13].

Body mass index (BMI) was calculated using the standard formula of weight in kilogram divided by square of height in meters (kg/m²).

By univariate analysis a number of variables including the factors mentioned by ADA as risk factors for type II diabetes [13] in addition to status and etiology of liver disease, interferon therapy, and history of or ongoing habitual smoking were compared between diabetic and non-diabetic patients (Table ​(Table1).1). We found that the prevalence of DM was significantly associated with older age (P = 0.000), CLD status (P = 0.000), history of hypertension (P = 0.007), family history of DM (P = 0.000), and higher BMI (0.009). Excluding the inactive carriers, we compared DM rate in two other groups. The prevalence of DM among cirrhotic cases (53.7%) was significantly higher than that of chronic hepatitis patients (13.7%), (P = 0.000).

At multivariate analysis age (P = 0.008, OR = 4.7, 95%CI: 1.8–12.2), family history of DM (P = 0.000, OR = 6.6, 95%CI: 2.6–17.6), chronic hepatitis (0.000, OR = 11.6, 95%CI: 2.9–45.4), and cirrhosis (P = 0.000, OR = 6.5, 95%CI: 2.4–17.4) showed up as the only factors independently associated with DM prevalence rate (table ​(table22).

Since both cirrhosis and chronic hepatitis were found as independent factors associated with DM occurrence, we analyzed these two groups separately. The results were summarized in table ​table33.

Out of 41 cirrhotic patients, 22 (53.7%) patients were diabetic and 7 (17.1%) patients had IFG and/or IGT. Univariate analysis showed that only two factors were associated with DM rate. Mean of age in diabetic cases was significantly higher than that of non-diabetic ones (P = 0.003). Moreover, DM was significantly more prevalent in patients with Child-Pugh's score B than score A (P = 0.04). Since there was no score C in study sample, we could not evaluate this class of cirrhosis. After applying logistic model, only Child-Pugh's score kept its significance (P = 0.04, OR = 9.6, 95%CI: 1.0–88.4) as an independent predictive factor for DM in cirrhotic patients (table ​(table44).

Literature frequently demonstrated the higher prevalence of DM and IGT in cirrhosis than in chronic hepatitis [4,5,11]. However, about the higher DM rate in patients with chronic hepatitis compared with people without liver disease, there is not any widely general agreement. Some studies claimed that DM rate is not appreciably different when compared with general population or individuals without liver disease [11,17]. These studies believed that only cirrhosis but not chronic hepatitis were associated with DM [17]. In our study, multivariate analysis indicates independent association between chronic hepatitis and DM rate, despite the fact that we compared DM occurrence among three groups who all suffered from liver disease. Moreover, our data indicate that the frequency of DM increases significantly with the severity of the liver disease both in cirrhotic cases and in patients with chronic hepatitis. These findings suggest that liver fibrosis but not cirrhosis itself, is the event associated with glucose intolerance. A weak association between glucose intolerance and severity of liver disease in cirrhosis has already been reported by Muller et al [7], even though they used different criteria to evaluate liver damage severity. Another study applying Child-Pugh's classification showed similar results [11]. In this context it is worthy to note that although the underlying mechanism of glucose intolerance in cirrhosis has not been fully elucidated, it can be mainly explained by the insulin resistance [9], in addition to reduced glucose sensitivity in liver cirrhosis [18,19]. More interestingly, a recent study suggests that insulin resistance occurs already in the early stages of the chronic hepatitis course [12], and another study investigating chronic hepatitis patients with normal glucose tolerance revealed a strong relationship between insulin sensitivity and fibrosis score [20]. In contrast, there is another hypothesis that indicates CLD as a consequence of DM. This theory states that occurrence of insulin resistance initially facilitates lipolysis, and increases free fatty acid deposition in liver, which increases products of lipid peroxidation inducing oxidative stress. This results in cytokine-mediated hepatic inflammatory damage that induces collagen deposition and eventually fibrosis [21,22]. Although in this study we could not determine if the onset of DM was before or after of liver disease, our findings show that chronic hepatitis per se has an independent association with DM, and development of DM in chronic hepatitis patients was correlated to the severity of liver fibrosis.

No differences in DM rate were found when we studied patients according to CLD etiological categories both in cirrhosis and in chronic hepatitis group. A positive link between hepatitis C infection and development of DM has been suggested in some studies but not completely characterized. Mason et al surveying a large cohort of patients with viral chronic hepatitis showed a relatively strong association between DM and HCV infection [23]. They suggested that HCV infection may serve as an additional risk factor for the development of DM beyond that attributable to CLD alone [23]. In contrast, some studies provided evidence against a potential association between these two disorders [17,24]. A more recent study showed that the prevalence of DM among cirrhotic patients with hepatitis C was significantly higher than among those with cirrhosis due to cholestatic liver disease, but this rate difference was not significant when compared with cases with alcohol-induced cirrhosis [10]. In spite of the fact that they had not enrolled HBV infected cases in their sample population, it suggested that the mechanism of the DM in cirrhotic patients was related more closely to the underlying cause of liver cirrhosis [10]. Although no single etiological factor was found linked to DM occurrence in our study, we believe that our study was not designed to address this issue considering the fact that we had to exclude patients with NASH as another etiology of CLD because of small number of patients in this group. However, further prospective studies may shed more light on the relationship between DM and the underlying liver disease.

BMI did not differ between diabetic and non-diabetic cases among cirrhotic patients. This finding was expectable as reduction in the muscle mass is well described in liver cirrhosis, which can bias the comparison. On the other hand, among patients with chronic hepatitis BMI did remain as an independent predictive factor of DM. Some studies have indicated that obesity could be a potential risk factor for fibrosis in chronic hepatitis [25]. Our results suggest that increased BMI also has a role in the pathogenesis of DM in chronic hepatitis independent of liver fibrosis. Furthermore, Muller et al reported that basal free fatty acids and basal free glycerol plasma concentrations were increased in diabetic patients with liver cirrhosis when compared with those without diabetes [7]. In present study, TG and Chol levels in patients with chronic hepatitis and TG level in cirrhotic patients were higher in diabetic cases compared with non-diabetic ones although not statistically significant (table ​(table3).3). These findings may have therapeutic implications in the management of patients with chronic hepatitis. It appears necessary to greatly encourage overweight patients to make a concerted effort to lose weight in order to both decrease the risk of DM development and to prevent the liver damage.

Although family history of DM is a well-known risk factor for DM, the correlation between DM and both cirrhosis and chronic hepatitis remained significant even when family history of DM was entered through logistic model. This finding, in concordance with similar studies [7,11], indicates that liver injury per se is associated with DM and a family history of DM is only an adjunctive factor. However, the possible reporting bias of family history of DM should be carefully considered as it is not unreasonable to think that patients with known DM may be more likely to know or think that they have a family history of DM. When cirrhosis and chronic hepatitis groups were analyzed separately, we observed that patients with a positive family history of DM did not show an increased frequency of DM, particularly in cirrhotic patients. Slightly less than 45% of cirrhotic cases with a negative family history of DM were diabetic. This finding does not support the speculation that cirrhotic patients only with a genetic predisposition for DM are prone to glucose intolerance as the manifestation of their liver disease.

Age is another definite risk factor for type II DM in the normal population [13], and it was expected also to be associated with DM in CLD. However, the odds ratio for both cirrhosis and chronic hepatitis was higher than that for age, implying a stronger association of the former factors. Although in the part of our study investigating cirrhotic patients, age lost its significance when it was entered in multivariate analysis, the majority of studies demonstrated that IGT and also DM were more frequently seen at advanced age in cirrhotic patients [2,7,11].

While DM was associated with hypertension in univariate analysis (table ​(table1),1), in multivariate analysis it lost its significance as a predictor of DM (table ​(table2).2). These two variables may both be related to the metabolic syndrome and adjustment for other related variables removes this association. On the other hand, it is currently believed that cardiovascular disease is rare in cirrhosis and studies demonstrate that cirrhotic patients, even in the presence of overt DM, have a low prevalence of vascular disease including hypertension [26]. Our findings support this hypothesis showing no difference in the rate of hypertension between diabetic and non-diabetic patients.