Metformin in the treatment of patients with non-alcoholic steatohepatitis
Dr A. Kadayifci, Fevzi Cakmak Bulv. Kardelen apt. 86/10, Sehitkamil/Gaziantep, Turkey.
Background : Increased insulin resistance is the major pathogenic mechanism in the development of non-alcoholic steatohepatitis.
Aim : To investigate the therapeutic effect of metformin, a well-known insulin-sensitizing agent, in the treatment of non-alcoholic steatohepatitis.
Methods : Thirty-six patients with non-alcoholic steatohepatitis were randomized into two groups. The first group was given lipid and calorie-restricted dietary treatment alone, and the second group was given metformin 850 mg b.d. plus dietary treatment, for 6 months. The changes in biochemical, sonographic and histological parameters were compared.
Results : The mean serum alanine/aspartate aminotransferase, insulin and C-peptide levels decreased and the index of insulin resistance improved significantly from baseline in the group given metformin. The mean changes in these parameters in the metformin group were significantly greater than those in the group given dietary treatment alone. Although more patients in the metformin group showed improvement in the necro-inflammatory activity, compared with the group given dietary treatment alone, no significant differences in necro-inflammatory activity or fibrosis were seen between the groups.
Conclusion : The data suggest that improvement of the insulin sensitivity with metformin may improve the liver disease in patients with non-alcoholic steatohepatitis.
Non-alcoholic steatohepatitis is a common disease that may progress to end-stage liver disease. It is now considered to be a widespread liver disease in Western countries.1 Despite its common occurrence, there is no proven pharmaceutical therapy for patients. Without doubt, weight loss is the therapy of choice for those who are overweight, but this requires a dramatic change in dietary habits and lifestyle.1, 2 Therefore, it can be achieved or maintained by only a limited number of patients. Several drugs, such as clofibrate, gemfibrozil, ursodeoxycholic acid and some antioxidants, have been used in various studies for the treatment of non-alcoholic steatohepatitis.3–6
Ongoing studies and current reports have focused on the role of hyperinsulinaemia and increased insulin resistance as major pathogenic factors in the development of non-alcoholic steatohepatitis.7, 8 On the basis of these data, it has been suggested that a decrease in insulin resistance with drugs such as thiazolidinediones and metformin may be of therapeutic value.1, 9 Troglitazone produced some improvement in non-alcoholic steatohepatitis in one study, but was withdrawn from clinical use because of its hepatotoxicity.9 Recently, Neuschwander-Tetri et al. reported their interim results demonstrating the effects of rosiglitazone in patients with non-alcoholic steatohepatitis.10 Rosiglitazone improved insulin sensitivity, reduced liver fat content and improved liver enzyme concentrations in that study. Lin et al. showed that metformin reduced significantly the hepatomegaly and hepatic steatosis associated with insulin resistance in ob/ob mice.11 Moreover, Marchesini et al. showed that metformin reduced serum transaminase concentrations and liver volume in a recent study performed on humans for the first time.12 However, these results attributed to metformin have not been confirmed by any randomized studies. Furthermore, the effect of metformin treatment on human liver affected by non-alcoholic steatohepatitis has not been shown histologically. Metformin, apart from being non-hepatotoxic, is also cheaper and therefore more suitable for long-term treatment. In this first randomized controlled study, we investigated the effect of metformin on the insulin resistance, liver enzyme concentrations and hepatic histology of patients with non-alcoholic steatohepatitis.
From August 1999 to June 2001, a total of 36 patients with non-alcoholic steatohepatitis were enrolled in the study. To confirm the diagnosis of non-alcoholic steatohepatitis and to rule out other possible liver diseases, all patients underwent a detailed clinical and laboratory evaluation, including liver enzyme concentrations, hepatitis markers [hepatitis B surface antigen (HBsAg), antibody to hepatitis B core antigen (anti-HBcIgG), anti-hepatitis C virus (anti-HCV)] and autoantibodies (anti-nuclear antibody, anti-mitochondrial antibody, anti-smooth muscle antibody). Patients with suspected acute or chronic viral hepatitis, autoimmune hepatitis or any other liver disease were excluded. The exclusion criteria were as follows: relative or absolute contra-indication for metformin, possible liver disease other than non-alcoholic steatohepatitis, history of malignant disease, impaired renal function (serum creatinine > 1.5 mg/dL), heart failure, history of lactic acidosis, severe infection, hypoxic status, serious acute and chronic illnesses, homodynamic instability, age more than 70 years, diabetes mellitus or the current use of any drugs that may affect the results. Alcohol consumption was denied in 27 patients and negligible (< 100 g of ethanol per month) in the remaining nine patients. The patient's spouse or one of the family members was also asked about the alcohol intake of the patients. Patients with elevated γ-glutamyl transpeptidase activity of more than 75 U/L (normal range, 5–50 U/L) were excluded from the trial as a result of possible concealed alcohol intake. Patients fulfilling the criteria described by the National Cholesterol Education Expert Panel (ATPIII) were also defined as having metabolic syndrome.13
The study intended to: (i) clarify the effect of metformin on insulin resistance in patients with non-alcoholic steatohepatitis; (ii) determine whether metformin improves biochemical and histological parameters of liver damage in patients with non-alcoholic steatohepatitis; (iii) establish the superiority of treating non-alcoholic steatohepatitis with metformin plus diet vs. diet alone. After the selection procedure, patients were randomly assigned into two study groups using random sampling numbers. The first group of patients was given dietary treatment alone and the second group of patients was given metformin 850 mg b.d. plus dietary treatment, for 6 months. For dietary treatment, a dietician was consulted by all patients to restrict their intake of lipids and non-complex carbohydrates. All obese and overweight patients were advised to lose weight with a restriction of daily calorie intake to 1600–1800 calories per day. The compliance of patients to their diet was evaluated using a questionnaire during the follow-up period and classified as poor, moderate and good. Informed written consent was obtained from each subject and the local ethical committee approved the study. All patients were evaluated on an out-patient basis.
Serum insulin, fasting glucose and C-peptide levels were analysed, in addition to other biochemical tests, such as liver tests and serum lipids, in all patients. Blood samples were collected in the morning after an overnight fast, and analysed on the same day. The index of insulin resistance was calculated on the basis of fasting values of plasma glucose and insulin, according to the homeostasis model assessment method [insulin resistance (%) = insulin/22.5 × e– ln(glucose)].14 Blood samples were also obtained from patients in the metformin group to assess plasma lactic acid levels before the study. Plasma lactic acid levels were determined by the dry-slide method using an automated analyser (Vitros DT II, Johnson & Johnson Company, NY, USA). The reference interval was 0.7–2.1 mmol/L. The body mass index was calculated in all patients and obesity was defined as a body mass index greater than 30 kg/m2. Hyperlipidaemia was defined as a serum total cholesterol level more than 220 mg/dL or triglyceride level more than 200 mg/dL.
All subjects underwent an upper abdominal ultrasonography before the study. The detection of a bright, diffusely echogenic liver was accepted as concordant with hepatic steatosis. The pattern and severity of the increase in echogenicity and liver volume in upper abdominal ultrasonography were graded as mild, moderate or severe.15 Non-alcoholic steatohepatitis was also confirmed in all patients by histopathological examination following liver biopsy. The biopsy was performed percutaneously using a Tru-Cut needle with ultrasound guidance by a gastroenterologist, and a sample length of at least 20 mm was regarded as an acceptable specimen. The combination of hepatocellular steatosis, ballooning and disarray, and acinar or portal inflammation in histopathological examinations was graded according to the necro-inflammatory activity grading system for steatohepatitis (grade 1, mild; grade 2, moderate; grade 3, severe).16 The type and progression of fibrosis observed in the liver biopsies were staged as proposed by Brunt et al.16 According to this scoring system, the fibrosis pattern in the liver was defined as follows: stage 0, no fibrosis; stage I, zone 3 fibrosis; stage II, zone 3 and periportal fibrosis; stage III, bridging fibrosis; stage IV, cirrhosis. The presence of fibrosis (grade I or more), or grade 2–3 necro-inflammatory activity in the case of no fibrosis, represented the minimal histological criteria used to define non-alcoholic steatohepatitis.
The patients in the metformin group were contacted by telephone in the first and second weeks of treatment and questioned about any side-effects attributed to the drug. They were also warned of possible hypoglycaemia and gastrointestinal side-effects due to metformin. In both groups, all serum parameters (and, in the metformin group, the serum lactate level) were repeated at the end of the first, third and sixth months of treatment. Abdominal sonography and liver biopsy were also repeated at the end of treatment in patients in whom informed consent had been obtained. In both groups, the patients were advised to maintain their dietary recommendations after the study period and were recalled for clinical evaluation and liver enzyme concentrations every 3 months.
All results were expressed as the mean ± s.d. All mean serum parameters and other variables were compared before, during and after treatment in the two groups by Kruskal–Wallis test for variance analyses and by Wilcoxon signed rank test for dual comparisons. The Mann–Whitney U-test and chi-squared test were used to compare the differences in demographic characteristics and other variables between the groups. All analyses were two-tailed and were conducted using computer-based statistics software (SPSS for Windows 8.0, 1997, SPSS Inc., Chicago, IL, USA). A P value of less than 0.05 was accepted as statistically significant.
Thirty-four patients completed the study protocol successfully. One patient in the group given dietary treatment alone did not come to control visits regularly, and one patient in the metformin group was excluded from the study because of a suspicion of autoimmune hepatitis that appeared at the third month. The demographic characteristics of the groups are summarized in Table 1. In both groups, nearly one-half of patients reported good compliance with their dietary recommendations, whereas the compliance was moderate or poor in the others. No significant difference in dietary compliance was observed between the two groups. The co-morbidities associated with non-alcoholic steatohepatitis in the group given dietary treatment alone were as follows: obesity in five, hyperlipidaemia in three, obesity and hyperlipidaemia in seven and undefined in two. In the metformin group, the associated co-morbidities were as follows: obesity in six, hyperlipidaemia in two, obesity and hyperlipidaemia in eight and undefined in one.
Table 1. Demographic characteristics of the patients
|Number (n)||17||17|| |
|Age (years, range)||41.5 ± 9.1 (23–61)||39.8 ± 10.6 (22–64)||0.12|
|Body mass index (kg/m2)||28.4 ± 3.9||30.1 ± 3.4||0.09|
The results of the mean serum parameters in both groups during the study period are shown in Tables 2 and 3. A significant decrease was noted in the mean serum alanine aminotransferase, aspartate aminotransferase, insulin and C-peptide levels and in the index of insulin resistance during and at the end of treatment in the metformin group. The rate of decrease was found to be most prominent during the first month of treatment, and continued to decrease during the study period. Comparison of the mean serum alanine aminotransferase and aspartate aminotransferase levels and the index of insulin resistance before and after treatment showed a strong statistical difference in the metformin group (P < 0.0001). A gradual mild decrease was observed in the mean serum glucose, total cholesterol and triglyceride levels during treatment without reaching statistical significance (P > 0.05). The mean serum alanine aminotransferase and aspartate aminotransferase levels also decreased in the group given dietary treatment alone (P = 0.001 and P = 0.06, respectively), but was not as significant as in the metformin group. The other serum parameters, including insulin and C-peptide levels and the index of insulin resistance, did not change significantly in the dietary treatment group compared with the mean levels before treatment.
Table 2. Results of biochemical parameters in the group given dietary treatment alone (mean ± s.d.)
|Alanine aminotransferase (U/L)||72.8 ± 31.2||70.6 ± 28.5||62.7 ± 24.5||55.4 ± 16.3||0.001|
|Aspartate aminotransferase (U/L)||48.1 ± 26.3||46.3 ± 25.8||45.5 ± 17.8||41.3 ± 13.5||0.06|
|Glucose (mg/dL)||96.4 ± 19.2||97.3 ± 18.8||95.2 ± 17.3||92.7 ± 15.7||0.04|
|Body mass index (kg/m2)||28.4 ± 3.9||28.1 ± 4.1||26.2 ± 3.2||26.5 ± 3.7||0.01|
|Insulin (µIU/mL)||13.7 ± 4.7||13.2 ± 4.9||13.2 ± 5.1||11.8 ± 4.4||0.05|
|C-peptide (ng/mL)||4.95 ± 5.6||5.1 ± 4.9||4.87 ± 5.2||4.86 ± 5.8||0.09|
|Index of insulin resistance (%)||1.83 ± 0.74||1.82 ± 0.81||1.85 ± 0.72||1.81 ± 0.67||0.18|
|Cholesterol (mg/dL)||197.8 ± 39.9||191 ± 40.1||180.1 ± 35.6||173.4 ± 31.7||0.01|
|Triglyceride (mg/dL)||203 ± 68.5||188 ± 54.2||184.4 ± 49.7||185.3 ± 51.6||0.13|
Table 3. Results of biochemical parameters in the group given metformin and dietary treatment (mean ± s.d.)
|Alanine aminotransferase (U/L)||83.5 ± 24.6||70.2 ± 25.3||61.6 ± 35.9||46.4 ± 23.3||0.0001|
|Aspartate aminotransferase (U/L)||57.9 ± 17.3||49.1 ± 14.6||39.8 ± 12.8||35.8 ± 10.5||0.0001|
|Glucose (mg/dL)||87 ± 14.2||86.5 ± 13||84.8 ± 8.1||80.7 ± 6.7||0.033|
|Body mass index (kg/m2)||30.1 ± 3.4||29.3 ± 3||28.4 ± 2.7||27.7 ± 2.5||0.001|
|Insulin (µIU/mL)||12 ± 5.1||10.4 ± 4.6||9.06 ± 4.3||7.1 ± 4||0.001|
|C-peptide (ng/mL)||4.18 ± 1.3||3.7 ± 1.4||3.5 ± 1.3||2.8 ± 1.3||0.003|
|Index of insulin resistance (%)||2.53 ± 0.98||2.2 ± 0.9||1.86 ± 0.76||1.38 ± 0.71||0.0001|
|Cholesterol (mg/dL)||189 ± 39.2||184 ± 31||182 ± 24.7||171 ± 50||0.05|
|Triglyceride (mg/dL)||178 ± 70.5||163 ± 74.1||160 ± 70||172 ± 65.8||0.68|
|Lactic acid (mmol/L)||1.78 ± 0.4||2.07 ± 0.48||1.98 ± 0.0.41||2.11 ± 0.52||0.21|
For both groups, the mean (± s.d.) change in the serum parameters, index of insulin resistance and body mass index between time zero and the sixth month of treatment and the percentage rate are shown in Table 4. At the end of the sixth month, the serum alanine aminotransferase level had reached the normal range in 10 of 17 patients (59%) and the aspartate aminotransferase level in 12 of 16 patients (75%) in the metformin group, whereas the corresponding values were six of 16 patients (37.5%) and four of 15 patients (26%), respectively, in the group given dietary treatment alone. The rate of decrease of the mean serum insulin and C-peptide levels and the index of insulin resistance was also more significant in the metformin group than in the group given dietary treatment alone (P < 0.05). The body mass index decreased by more than 1 point in 10 patients in the metformin group and in seven patients in the dietary treatment group. The mean body mass index had decreased significantly in both groups by the end of the sixth month, but the results were more prominent in the metformin group. Thus, metformin plus dietary treatment was observed to be more effective for the achievement of weight loss than dietary treatment alone. The basal level of serum lactic acid increased by more than 0.5 mmol/L in six patients and was elevated above the upper limit in three patients in the metformin group during treatment, but no signs or symptoms of lactic acidosis were detected. The mean increase in the serum lactic acid level was not statistically significant at the end of treatment (Table 4). Fourteen patients in the metformin group (82.5%) and 13 patients in the dietary treatment group (76.5%) had three of the five components of the metabolic syndrome according to ATPIII criteria. No significant difference was observed in the response to treatment in patients with or without metabolic syndrome in both groups. As the number of cases without metabolic syndrome was insufficient, no statistical comparison was performed.
Table 4. Mean (± s.d.) changes in parameters and percentage rate at the end of treatment in the two groups
|Alanine aminotransferase (U/L)||17.4 ± 14.1 (24)||37.1 ± 22.2 (44.4)||0.003|
|Aspartate aminotransferase (U/L)||6.8 ± 5.9 (14)||22.1 ± 14.3 (38)||0.0001|
|Glucose (mg/dL)||3.7 ± 7.1 (3.8)||6.3 ± 11.2 (7.2)||0.38|
|Body mass index (kg/m2)||1.9 ± 2.1 (6.7)||2.4 ± 1.9 (7.9)||0.01|
|Insulin (µIU/mL)||1.9 ± 0.9 (13.8)||4.9 ± 3.7 (40.8)||0.002|
|C-peptide (ng/mL)||0.09 ± 0.2 (1.8)||1.4 ± 1.3 (33)||0.002|
|Index of insulin resistance (%)||0.02 ± 0.03 (1.1)||1.15 ± 0.82 (45)||0.001|
|Cholesterol (mg/dL)||24.4 ± 41.1 (12.3)||18.4 ± 37.6 (9.7)||0.9|
|Triglyceride (mg/dL)||17.7 ± 39.4 (8.7)||6.3 ± 48.4 (3.5)||0.8|
|Lactic acid (mmol/L)|| ||0.33 ± 0.2 (18.5)|| |
Four patients in the metformin group and seven patients in the dietary treatment group refused the control biopsy even though written consent had been obtained at the beginning of the study. For this reason, control liver biopsies were performed in 13 patients in the metformin group and in 10 patients in the dietary treatment group at the end of treatment. The mean histopathological scores of necro-inflammatory activity and fibrosis before and after treatment are summarized in Table 5. There was a slight decrease in the mean grade of necro-inflammatory activity in the metformin group, but this result was not statistically significant. No change in the mean grade of necro-inflammatory activity was detected in the group given dietary treatment alone. The grade of necro-inflammatory activity was unchanged in five patients, improved in six and had progressed in two in the metformin group, whereas it was unchanged in seven, improved in none and had progressed in three in the dietary treatment group. The fibrosis score was unchanged in 10 patients, improved in none and had progressed in three in the metformin group, whereas it was unchanged in eight, improved in none and had progressed in two in the dietary treatment group. The frequency of improvement in the necro-inflammatory activity was greater in the metformin group than in the group given dietary treatment alone (46% vs. 10%), but the result was not statistically significant (P = 0.17 by Fisher's exact test).
Table 5. Mean histopathological necro-inflammatory activity grade and fibrosis score in the two groups before and after treatment
|Necro-inflammatory activity||1.41 ± 0.6||1.3 ± 0.48||0.62||1.41 ± 0.61||1.15 ± 0.68||0.31|
|Fibrosis||1.05 ± 1.1||1.12 ± 1.1||0.91||0.94 ± 1.02||0.92 ± 1.03||0.96|
The mean grade of steatosis by upper abdominal ultrasonography decreased from 1.53 ± 0.73 to 1.28 ± 0.63 in the dietary treatment group and from 1.62 ± 0.51 to 0.98 ± 0.43 in the metformin group at the end of the sixth month (P = 0.17 and P = 0.038, respectively). The sonography grade showed improvement in nine patients (53%) and was evaluated as normal in five of these patients (29%) at the end of therapy in the metformin group. A decrease in sonography grade was only detected in three patients (17.6%) and was unchanged in the others in the dietary treatment group.
None of the patients discontinued metformin because of intolerability during treatment. No patient reported symptoms of hypoglycaemia. Four patients complained of gas and bloating and two patients of mild to moderate abdominal pain in the first month. However, these complaints did not require cessation of the drug.
A 6-month follow-up period in 15 patients and a 1-year follow-up period in 11 patients were completed after metformin treatment. The mean serum alanine aminotransferase levels at 6 and 12 months of follow-up were 45.1 ± 19.9 U/L and 44.2 ± 16.7 U/L, respectively. The mean aspartate aminotransferase levels were 37.8 ± 13.9 U/L and 39.4 ± 12.8 U/L, respectively. The body mass indices were 27.1 ± 3.1 kg/m2 and 26.7 ± 3.9 kg/m2 at 6 and 12 months of follow-up, respectively. No significant change in the mean alanine aminotransferase and aspartate aminotransferase levels or body mass index was detected during the follow-up period. Serum alanine aminotransferase and aspartate aminotransferase levels were elevated above the upper limits in only one patient at follow-up, but normalized in two other patients.
Metformin increases insulin-mediated glucose utilization in peripheral tissues and has an anti-lipolytic effect that lowers serum free fatty acid concentrations.17 In this study, metformin was found to cause a decrease in serum insulin and C-peptide concentrations by producing a significant improvement in insulin action. This improvement probably led to a significant reduction in the accumulation of free fatty acids in hepatocytes and suppressed the oxidation of fatty acids contributing to cell injury and inflammation. A significant and greater decrease in the mean serum alanine aminotransferase and aspartate aminotransferase levels in the metformin group than in the group given dietary treatment alone confirmed that the drug probably restricted the damage to hepatocytes. The biochemical response to treatment continued during the study period, suggesting a sustained effect of the drug. In this study, the effect of metformin treatment on hepatic histopathology was evaluated for the first time in patients with non-alcoholic steatohepatitis. More patients in the metformin group than in the dietary treatment group showed improved hepatic necro-inflammatory activity, with decreased steatosis, ballooning of hepatocytes and acinar/portal inflammation, but the difference between the groups was not significant. No significant change was detected in the fibrosis score of the liver at the end of treatment in both groups.
Solomon et al. showed that metformin reversed the insulin resistance induced by tumour necrosis factor-α in liver cells.18 Lin et al. also suggested that the potential mechanisms of metformin with regard to the elimination of fat from the liver in animal models were probably related to the inhibition of hepatic tumour necrosis factor-α and of several tumour necrosis factor-inducible responses promoting hepatic steatosis and inflammation.11 In a recent study, Zhou et al. showed that metformin regulated the adenosine monophosphate-activated protein kinase in hepatocytes, which is a major cellular regulator of lipid and glucose metabolism.19 These authors suggested that the activation of the adenosine monophosphate-activated protein kinase provided a unified explanation for the beneficial effect of metformin. All of these studies have attempted to explain the mechanism of metformin in the liver and have reached a consensus that metformin may prevent, restrict or reverse hepatic steatosis and inflammation in non-alcoholic fatty liver disease.
Older age, obesity and the presence of diabetes mellitus have been reported to be independent predictors of more advanced disease in patients with non-alcoholic steatohepatitis by Angulo et al.20 In addition, Marchesini et al. have shown that the frequency of metabolic syndrome based on ATPIII criteria is high (88%) in patients with non-alcoholic steatohepatitis, and these patients carry a higher risk of potentially progressive liver disease.21 Using the same criteria, we also found a high rate of metabolic syndrome in patients with non-alcoholic steatohepatitis. This is an extremely important point, as this sub-group of patients would be expected to derive more benefit from metformin treatment.
Lactic acidosis is the most important potential risk during metformin treatment. The US Food and Drug Administration reported about five cases of lactic acidosis in 100 000 metformin-treated patients in a year.22 In a recent study, Lalau and Race searched the medical reports for a link between lactic acidosis and metformin treatment.23 They found no true metformin-associated lactic acidosis or mortality due to metformin alone. It seems that lactic acidosis is always associated with other contributing factors. In this study, the appearance of autoimmune hepatitis in one patient during the third month of therapy raised the suspicion of an adverse event of metformin. However, we do not have sufficient data to make a clear comment.
The small number of patients, the unblind nature of the study and the lack of a placebo group were major drawbacks of this investigation. In addition, post-treatment liver biopsy could not be performed in a certain proportion of patients. Moreover, the treatment period was short and a longer course may be necessary to observe more significant changes in hepatic histology. The baseline fibrosis score was low in the study because some patients without fibrosis, but with moderate ballooning and chronic inflammation in biopsies, were included. This might have limited the overall impact of the study, as the patients most likely to benefit are those with more advanced stages of fibrosis.
In conclusion, 6 months of therapy with metformin was well tolerated by patients with non-alcoholic steatohepatitis and led to a greater improvement in insulin resistance and liver enzymes than dietary treatment alone. Although metformin seemed to improve the severity of steatohepatitis in some patients, no effect on fibrosis was seen. The results of this pilot study suggest that metformin should be evaluated in larger controlled trials with extended follow-up and liver histology as the end-point.