Journal Logo

The Liver, Liver Disease, and Diabetes Mellitus

Albright, Eric S. MD*; Bell, David S. H. MB, FACE†

Author Information
  • Free
  • Take the CME Test


The insulin receptors in the liver serve as the treatment target in patients with diabetes or prediabetes by regulating gluconeogenesis, glycogenolysis and hepatic glucose production. Diabetes mellitus (DM) is a disease that affects multiple organ systems and persistent hyperglycemia can result in damage to noninsulin sensitive organs where there are no “gate keepers” in the form of insulin receptors restricting the entry of glucose into the cell. The insulin receptors in the liver may spare it direct damage from hyperglacemia. Although hyperglycemia does not cause liver damage, liver disease occurs at a higher frequency in diabetic patients and some oral hypoglycemics can cause liver damage. Hepatic diseases that have been observed more frequently with diabetes include: nonalcoholic fatty liver disease, nonalcoholic hepatic steatosis, nonalcoholic steatohepatitis (NASH), hemochromatosis, viral hepatitis, hepatic autoimmune disease, cholelithiasis, and primary liver cancer.

Increased prevalence of liver disease occurs in both type 1 (DM1) and type 2 (DM2) diabetic patients, resulting in an increased prevalence of hepatic complications including: cirrhosis, portal hypertension, liver failure, steatosis, iron overload, and hepatoma. Therefore, endocrinologists have found themselves treating more patients with liver disease because of the above association and the increased incidence of insulin resistance, glucose intolerance, and diabetes mellitus in patients with primary liver disease.


The hepatitis C virus (HCV) has infected nearly 4 million Americans and of these subjects 85% develop chronic active hepatitis and 20% eventually develop cirrhosis [1–3]. With the high prevalence of HCV in the general population it is not surprising that many subjects with HCV have diabetes. However, HVC, but not HVB, has been shown to have a higher prevalence in the DM2 patient [1].

How strong is the association of hepatitis C and DM2? The Third National Health and Nutrition Examination Survey (NHANES) showed that of the 984l patients older than 20, 8.4% had diabetes, and 2.1% had hepatitis C. This translated into a 3.7 fold increase in the risk for having or developing DM2 in hepatitis C infected patients [3]. Patients with hepatitis C who are older than 40 and continue to drink alcohol increase their risk for developing DM2 compared with those who stop drinking [3,4]. This study did not show an association between DM1 and HCV, nor an association between hepatitis B virus (HBV) and DM1 or DM2. The lack of an association with HBV is probably because HBV is confined to the liver, whereas HCV activity is more protean with multiple tissues being affected. In keeping with this theory it has been reported that a Canadian patient developed diabetes mellitus after a successful liver transplant for terminal HCV cirrhosis [5].

In a study by Mason et al [6], 21% of HCV patients and 12% of HBV patients were found to have diabetes after excluding medications that might induce hyperglycemia. The prevalence observed by Mason et al [6] was higher than that in the NHANES cohort, but rates similar to Mason et al [6] have been observed in studies performed in European and Middle Eastern countries [3,7,8]. Other studies have found HCV infection to be a statistically significant independent predictor for the development of diabetes mellitus [3,4,6]. Patients with cirrhosis were excluded from these analyses, and it has been shown that cirrhosis accompanied by HCV infection increases the risk for glucose intolerance and DM [9]. It has also been observed that patients with HCV infection have decreased first phase insulin release, which is characteristic of early beta cell dysfunction [8]. Insulin resistance is a common finding among all cirrhotic patients, but the incidence of diabetes among HCV positive cirrhotic patients is greater than the incidence of diabetes found among patients with cirrhosis caused by other etiologies [3,4,6]. The most prevalent HCV genotype in the United States is 1a or 1b, but it is the HCV 2a genotype that has been found to occur at a significantly higher frequency in diabetic patients [6]. The HCV 2a genotype is the genotype that is most associated with extra-hepatic manifestations of HCV infection, which is a possible explanation for its strong association with diabetes.

Increased prevalence of diabetes in the HCV patient is most likely caused by a direct infection of the pancreatic beta cell, but a destructive mechanism in response to the presence of the virus in the pancreatic tissue cannot be ruled out as a cause of beta cell dysfunction. The presence of an increased incidence of other autoimmune diseases, such as thyroiditis, thrombocytopenia, and lichen planus, that occur in HCV infected persons is evidence for an underlying autoimmune etiology [10]. Therefore, molecular mimicry causing an autoimmune response is a possible but less likely cause of HCV associated diabetes. In DM1 patients the autoimmune mechanism (in most cases) ultimately destroys the pancreas beta cell and because total beta cell destruction is rare in the HCV infected subject an autoimmune mechanism is less likely. Furthermore, there is minimal, if any, correlation of HCV with DM1. In addition, most HCV positive diabetic patients do not express anti-islet or anti-GAD antibodies [6]. Overall, it can be concluded that an association between HCV and DM2 exists based on the increased prevalence of DM 2 in patients with HCV [1,3,5], and an increased prevalence of HCV in patients with DM2 [6].


Nonalcoholic fatty liver disease (NAFLD) is the most common form of liver disease in the United States and its prevalence is estimated to be 5% in the general population [11]. NAFLD is strongly associated with DM2 and obesity but without liver biopsies the true prevalence is unknown. Less invasive ultrasound evaluation has demonstrated between 21% and 78% of diabetic patients have a “fatty liver” [11,12]. The more sinister form of hepatic steatosis is NASH, which occurs in l7% to 25% of patients with NAFLD. Nonalcoholic steatohepatitis (NASH) is characterized by perisinusoidal inflammation, hepatic necrosis, and regeneration [11,13]. Disruption of the liver architecture by NASH can therefore lead to cirrhosis, portal hypertension, and hepatic failure. Why only a small proportion of patients progress to NASH from NAFLD is not known. The characteristics of those most commonly affected are obesity, DM2, and female gender [11,14,15]. Progression is hypothesized to be a two-step process initiated by insulin resistance resulting in NAFLD. Later with the chronic increase in hepatic triglyceride there is stimulation of cytokines that results in hepatic inflammation, necrosis, cirrhosis, portal hypertension, hepatic failure, and death [11]. Tumor necrosis factor alpha (TNF α) may be the cytokine most commonly involved in this process and higher levels of TNF α are found in patients with the characteristics of the insulin resistance syndrome and in subjects with glucose intolerance [16]. Currently, the diagnosis of NASH is based on the application of strict pathologic criteria. Although the prevalence of DM2 in patients with NASH has been estimated to be anywhere from 2% to 55%, the degree of hepatic steatosis has been shown to be more closely linked to the presence of obesity than the presence of DM2 [12].

Nonalcoholic fatty liver disease is associated with all the classic features of the insulin resistance syndrome [12,17–20]. Patients with NAFLD have a significant increase in waist to hip ratio when compared with matched controls [17]. The NAFLD patients also had higher serum triglyceride, fasting insulin and C-peptide levels, and a lower HDL level. Although fasting glucoses were all within the normal range they were significantly higher in the group with NAFLD.

In a study group of DM2 patients, the best predictor of the amount of insulin required to reduce the serum glucose had a direct linear correlation to hepatic fat content [12,17,18]. This would suggest that hepatic fat content correlates directly with insulin resistance. Furthermore, if NASH rather than NAFLD is present there appears to be more insulin resistance [16]. A retrospective review of patients with biopsy proven NASH found that most patients had clinical signs and symptoms of the metabolic syndrome. Eight-five percent demonstrated insulin resistance on glucose tolerance testing suggesting that steatosis may be the hepatic expression of the metabolic syndrome [19].

In animal studies, thiazolidinediones have been shown to redistribute both peritoneal and hepatic fat to the subcutaneous space. Recent human data has also demonstrated fat redistribution with both currently available thiazolidinediones [20,21]. In the DM2 patients, unpublished studies of thiazolidinediones have demonstrated as much as a 30% reduction in hepatic fat utilizing ultrasonography and a significant reduction utilizing serum GGT as a marker for hepatic fat content.

Studies are now in progress to assess whether lowering of insulin resistance and hepatic fat with TZDs will result in stabilization or even an improvement of the steatohepatitis seen on liver biopsies of diabetic and nondiabetic patients. The use of a TZD will probably be a major advance in the therapy of the steatohepatic patient and TZDs will undoubtedly be utilized in nondiabetic patients with NAFLD and possibly NASH. The current therapies utilized for NASH include Vitamin E and other antioxidants, but the only medication showing benefit is ursodeoxycholic acid [11]. A pilot study using troglitazone in NASH patients demonstrated normalization of ALT in 70% of the patients with a modest improvement of histopathology on repeat liver biopsy [22]. It is unclear whether a longer duration of therapy would have shown greater improvement, or if one of the newer TZD formulations would have been more effective. However, these authors cautioned against utilizing ALT levels rather than liver histology as a marker of improvement in NASH patients. Larger studies utilizing the currently available TZDs for a longer duration will help clarify these questions. Furthermore, early intervention with TZDs in patients with steatohepatitis may lower the number of patients who progress to NASH. TZDs have both anti-inflammatory and anti-proliferative properties that could possibly decrease the number of patients who develop NASH. Recent ultrasonography data in a group of nondiabetic patients with steatohepatitis showed a reduction of liver volume by 20% with metformin therapy compared with 10% with diet therapy [23]. Because metformin use is contraindicated in patients with hepatic decompensation, it should be noted that even though these patients did not develop lactic acidosis, the lactic acid levels did increase [23]. Therefore, caution should be utilized in prescribing metformin if any degree of hepatic impairment is present. Larger studies are needed to determine the safety and efficacy of metformin in the treatment of steatohepatitis. Patients who have hepatic insufficiency should not use metformin, whereas patients with elevated transaminase levels should use metformin cautiously.


Hemochromatosis is an autosomal recessive disease, which causes parenchymal iron overload in both the liver and the pancreas. Hemochromatosis is an under-diagnosed yet treatable cause of both diabetes and liver disease. The gene for hemochromatosis is linked to the HLA locus on chromosome 6 and most homozygotes have a substitution in codon 282 where a cystine is replaced with a tyrosine [C282Y] [24]. The prevalence of heterozygotes is 1 in 10 among persons of European origin, and this has been confirmed in large observational studies.

A retrospective study revealed a high prevalence of hemochromatosis in European men diagnosed with DM1 after the age 30 [24]. An increased relative frequency of the homozygous C282Y mutation was found in these patients with late onset of DM1 compared with normal controls [24]. The theory of this relationship is that undiagnosed hemochromatosis will result in increased deposition of iron in the pancreatic beta cell leading to free oxygen radical formation and accelerated destruction of the beta cells [24]. Genetic screening for hemochromatosis would not be cost effective, but screening of patients with late onset DM1 with ferritin and iron saturation levels would be more appropriate.

To evaluate a possible link between DM2 and iron overload, Niederau et al [25] compared a small cohort of lean (BMI<25), noncirrhotic, idiopathic hemochromatosis (IHC) patients with matched healthy controls and demonstrated an increased prevalence of insulin resistance in the IHC patients. In this small cohort, despite increased pancreatic iron deposition, no differences were detected in alpha and beta cell secretory function [25]. However, a glucose load resulted in similar insulin release based on C-peptides, but the glucose peaks were higher in the IHC group [25]. Therefore, both insulin resistance and beta cell dysfunction were present. This suggests that the increased risk for DM2 in the IHC patient may be in part caused by pancreatic islet damage from iron deposition [25]. A recent review of l39 liver biopsies demonstrated a high degree of steatosis and inflammation in patients with iron overload, which may worsen hepatic insulin resistance and potentially accelerate pancreatic beta cell exhaustion [26]. Alternatively, postprandial hyperglycemia resulting from the lost first phase insulin release may stimulate a late and sustained elevation in plasma insulin levels, a down-regulation of insulin receptors, and insulin resistance. Therefore, it is probable that the mechanism for the increased prevalence of DM2 in patients with IHC is based on damage to the pancreatic beta cells [25]. As iron deposition occurs and destruction of the beta cells progresses, patients will pass through a period of insulin deficiency with or without insulin resistance. If beta cell destruction continues these patients will eventually require insulin.

There has been suspicion of a genetic link between the hemochromatosis gene (C282Y) and the genetic factors that predispose patients to DM2. A Canadian study found that 22% of DM2 patients had at least one copy of the C282Y mutation compared with 11.7% of the controls [27]. However, an English study found no difference in the incidence of the C282Y mutation between diabetic patients and controls [28,29]. Because the incidence of DM2 is much greater than that of hemochromatosis, screening of all patients with DM2 is not cost effective. If there is a genetic link between the two disorders and a genetic marker is identified, screening could be helpful. However, the larger English study sheds doubt that the presently identified gene can be used as a screening tool or even as a marker.

Hepatic iron overload that is not associated with hemochromatosis is referred to as primary iron overload (PIO). These persons tend to have normal transferrin saturation but an abnormal serum and a hepatic iron concentration that is disproportionately elevated when compared with a serum ferritin level [30]. A recent study has shown that PIO patients (similar to NAFLD and NASH patients) have features of insulin-resistance or the metabolic syndrome. Of these 65 patients studied with PIO, 72% had a body mass index (BMI) older than 25, 65% had dyslipidemia and 43% had glucose intolerance [30].

Why should PIO be associated with insulin resistance? There are three possible reasons. The first is that hepatic damage caused by PIO decreases the hepatic insulin extraction, elevating serum insulin levels causing a down-regulation of the insulin receptors. Second, is that the higher insulin levels associated with insulin resistance not only stimulates mobilization of glucose transporters, but also promotes mobilization of transferrin receptors (from the microsomal membranes) to the surface of the hepatocyte. There, the receptors can increase the uptake of iron and cause an excess accumulation of iron in the hepatocyte. Third, this inflammatory state associated with elevated cytokine levels may increase the transcription of ferritin messenger RNA in the macrophage. This increases iron uptake by these cells and the iron-laden macrophages could potentially act as a shuttle, transferring iron to the hepatocyte.

Whatever the mechanism, PIO is associated with the insulin resistance syndrome. Because NAFLD and NASH are also associated with the insulin resistance syndrome it is possible that the increased incidence of hepatic damage and cirrhosis associated with hepatic steatosis could be caused by primary iron overload and not the steatosis itself or vice versa. Therefore, this is an important relationship where early hepatic infiltration with either iron or fat results in hepatic insulin resistance which increases the risk for developing diabetes. The postulated link between the mutation for hemochromatosis and DM2 has yet to be determined, however in select DM1 patients who genetically are at high risk for hemochromatosis, screening with a ferritin level should be considered.


Autoimmune hepatitis (AIH) is a chronic form of hepatitis that occurs more frequently in women. It is characterized by derangements in the normal immune response and is associated with other autoimmune diseases. The clinical presentation can vary from asymptomatic inflammation to cirrhosis or fulminate hepatic failure.

Type 1 or classical AIH can occur at any age and is associated with antinuclear or antismooth muscle antibodies. The most common extrahepatic autoimmune manifestations are thyroid disease (Grave or Hashimoto thyroiditis), ulcerative colitis, and rheumatoid arthritis [31].

Type 2 AIH is seen most commonly in adolescent girls and young women and is associated with antibodies to liver/kidney microsomes (ALKM-1), or to liver cytosol antigen (ALC-1). There is a significant association of AIH with DM1, autoimmune thyroid disease, vitiligo, and 10% of patients with type 2 AIH have autoimmune polyglandular syndrome type 1 [APCED] [32,33]. Patients with APCED usually present as children with adrenal insufficiency, hypoparathyroidism and mucocutaneous candidiasis. They have a defect with the T-cell mediated immune axis and the syndrome has been linked to mutation of the autoimmune regulator gene on chromosome 21 [33]. The other common associations in addition to hepatitis are: dystrophy of dental enamel and the nails, vitiligo, and alopecia totalis. AIH type 2 patients need to be screened on a regular basis for DM1 with fasting blood serum glucose, C-peptides, and antiglutamic acid decarboxylase (GAD) antibodies and for the other autoimmune endocrine diseases mentioned above. There appears to be no known association of AIH 1 or 2 with the insulin resistance syndrome, nor is there an increased incidence of DM2.


Hepatogenous diabetes in patients with cirrhosis is a poorly understood term. Multiple studies have documented significant insulin resistance and glucose intolerance in patients with cirrhosis irrespective of the etiology and a lower, but a significant number of cirrhotic subjects has or develops frank diabetes [34,35]. Recent estimates suggest that 35% to 80% of all patients with cirrhosis have either impaired glucose tolerance or diabetes mellitus [35,36]. The proposed mechanisms for this finding are increased insulin resistance (mechanism is unknown), hepatic dysfunction itself, or coexistent pancreatic disease.

The hyperinsulinemia observed in cirrhosis is well recognized by both hepatologists and diabetologists, but the mechanism for this is unclear. Early studies did not demonstrate a difference in the insulin clearance rate based on C-peptide/insulin ratios in the steady state between cirrhotic patients and controls. During a glucose challenge the cirrhotic patient’s ratio of C-peptide to insulin fell suggesting deficient hepatic insulin uptake [37]. The decreased uptake of insulin by the liver during states of hyperglycemia has two probable mechanisms. First, the lack of insulin uptake will result in less inhibition of gluconeogenesis and hepatic glucose production and secondly it will decrease hepatic glucose uptake and storage.

An Italian study demonstrated low insulin secretion in cirrhotic subjects during both an oral glucose tolerance test and an intravenous glucagon challenge [38]. This is the only study that supports pancreatic insufficiency as a cause of glucose intolerance or diabetes associated with hepatic cirrhosis. These patients’ possible pancreatic insufficiency may have been confounded by the fact that they all had much further advanced hepatic disease compared with those in earlier studies.

In addition to the elevated insulin levels and the insulin insensitivity seen in cirrhotic patients, they also demonstrate a blunted first phase insulin release and possibly blunted second phase insulin release. It remains unclear what causes the progression to overt diabetes in a significant proportion of these patients. An intravenous glucose tolerance test demonstrated a lower glucose disposal rate in the cirrhotic groups with and without diabetes. The cirrhotic patients without diabetes had a larger insulin release and a longer duration of high plasma insulin levels compared with the cirrhotic group with diabetes [39,40]. These findings suggest that diabetes occurs in the insulin-resistant cirrhotic patient only when there is beta cell failure resulting in lower insulin release [41].

Treatment of liver failure with liver transplantation has demonstrated a reversal of insulin resistance and as much as a 67% cure of hepatogenous diabetes [42]. Liver transplantation improves both hepatic glucose clearance and peripheral glucose disposal by improving the insulin resistance associated with cirrhosis. The transplanted liver is more responsive to insulin allowing better regulation of glucose production, release, and storage; however, some residual hypersecretion of insulin remains with a glucose challenge [43]. Possibly the residual hyper-secretion of the insulin is based on a memory feature of the beta cell that would require time to down regulate, or the posttransplant immunosuppressive therapy may be causing mild insulin resistance. The patients who remained diabetic after transplantation all had beta cell dysfunction that persisted postoperatively.

Recently, a case control study demonstrated an increased incidence of both diabetes and obesity in patients with cryptogenic or idiopathic cirrhosis (CC) compared with patients with known causes of cirrhosis. DM2 was found in 47% of the CC group, which is four to five times the incidence in the general population of the United States [44]. The authors speculated that the cause of the CC was probably undiagnosed NASH, with the previously present fatty infiltrate being lost as hepatic damage increased and cirrhosis developed (which would explain the lack of histological findings on biopsy). This would suggest that the CC patients with hepatogenous diabetes probably had NASH as the cause of insulin resistance before the diagnosis of CC and also as the probable cause of their CC.

Whether hepatogenous diabetes exists, there is significant insulin resistance and an increased incidence of diabetes among cirrhotic patients. Although the cirrhosis itself plays a significant role, it is probable that other factors that trigger the development of diabetes are present. However, these factors must be reversible because up to 2/3 of hepatic transplant patients resolve their insulin resistance and diabetes.

Therefore, our hypothesis for the association of cirrhosis with insulin resistance and diabetes is based on a decreased extraction of insulin by the damaged liver. This leads to decreased hepatic glucose storage, higher serum glucose levels, and higher serum insulin levels. The need for more insulin production may cause beta cell decompensation with an initial loss of first phase insulin release and an exaggerated second phase insulin release that could potentially cause postprandial hypoglycemia. A continued increase in insulin resistance could finally result in diabetes. If loss of pancreatic function is not at an advanced stage the whole process could be reversed with a liver transplant.


Cholelithiasis is seen with a two-to-three fold increased frequency in both DM1 and DM2 patients compared with the general population [45]. Gallstones are formed when alterations in the chemical content of bile occur predisposing this supersaturated solution to form choleliths, and cholelith formation is enhanced with bile stasis. DM2 patients may have an increase in the cholesterol content of the bile and an increase in gallbladder volume creating a “primed” environment for stone formation [46,47]. Significant cholestasis can result from diabetic autonomic neuropathy. Although diabetic autonomic neuropathy induces cholestasis and the increased formation of gallstones, only 1/13 of these patients develop biliary colic, other obstructive effects, or cholecystitis. This is because autonomic denervation prevents full and efficient contraction of the gallbladder so that a gallstone is less likely to be propelled into and obstruct the bile ducts causing symptoms. Cholecystokinin has been hypothesized to play a role in impaired contraction of the gallbladder, but recent studies have demonstrated normal levels of cholecystokinin in DM patients [48,49]. The responsiveness of the gallbladder to physiological levels of cholecystokinin in DM patients compared with controls was not evaluated [45].

Based on the high frequency of asymptomatic cholelithiasis and higher operative complications in the diabetic patient, surgery is now recommended only for symptomatic patients [50,51]. Biliary cirrhosis caused by cholestatic liver disease has the same prevalence among patients with diabetes as it has in the general population, suggesting that the risk for acute or chronic cholestatic liver disease is not increased in the diabetic patient in spite of decreased gallbladder contractility and increased frequency of cholelithiasis [52].

Many studies have suggested a link between diabetes mellitus and liver cancer. The known risk factors for liver cancer include: excessive alcohol consumption, cirrhosis, HBV, HCV (more recently), autoimmune hepatitis, and hemochromatosis. A Swedish study found a three-fold increased incidence of liver cancer among DM2 patients after controlling for other risk factors [53]. The reason for the increased risk for liver cancer with DM2 was not delineated, but these observations occurred before HCV was classified. The incidence of HCV infection in DM2 is increased, and thus chronic liver disease caused by HCV may have caused the increased incidence of liver cancer in this Swedish study. DM2 did not prove to be an additional risk factor in a smaller study that evaluated for and confirmed HCV as a risk factor for liver cancer. This study was a Veterans Affairs chart review of patients with primary liver cancer and controls. In this evaluation, diabetes mellitus was associated with liver cancer only when present with established risk factors, especially alcoholic cirrhosis, HBV, and HCV [54]. Whether diabetes is an additive risk factor for hepatocellular carcinoma or an independent risk factor has yet to be determined. An increased vigilance is required in diabetic patients with any of the above listed risk factors and screening for hepatic tumors with ultrasonography should be performed on a regular basis.


Troglitazone was an effective hypoglycemic agent but had significant hepatotoxicity. Because of decreased exposure to troglitazone in clinical trials the true frequency of significant elevation of liver enzymes with troglitazone was not seen until after its release into the market [55].

After FDA approval a larger volume of patients were exposed to troglitazone during a longer period of time and hepatic deaths and liver transplants began to be reported. Practitioners were notified by mail of the possible hepatic risk with the recommendation to monitor liver function tests and to stop troglitazone if elevation of liver enzymes occurred. During the short period of availability of troglitazone, five to six different FDA label modifications occurred [56]. In spite of these warnings, at most, only 40% of patients exposed to troglitazone had their liver enzymes monitored on a regular basis [55]. Troglitazone because of its significant hepatotoxicity was eventually withdrawn from the US market.

Is the hepatotoxicity a class effect? Currently, at least 20% of prescriptions for oral hypoglycemics are for rosiglitazone and pioglitazone, and there is minimal evidence of hepatotoxicity. Three cases of hepatotoxicity in patients using rosiglitazone have been published but other potential causes of hepatotoxicity were present in these cases [57–59]. The first case reported was in a patient with severe congestive heart failure and hypotension who could have had “shock liver” from hepatic hypoperfusion [58]. The second patient was taking zafirlukast for asthma, which has been documented to increase the risk for hepatotoxicity, acute hepatic failure, and death [59]. The third case was an 82-year-old man who had been on therapy for greater than 1 year and had had normal liver enzymes a few months before admission. The only new medication exposure was testosterone gel; testing suggested acute hepatitis and liver failure, which was rapidly progressive and fatal [57]. We must conclude that rosiglitazone’s involvement in these cases of hepatotoxicity is questionable at best, at least in the first two cases. Pioglitazone has been reported to be associated with acute hepatotoxicity in two cases. The two cases involving pioglitazone occurred 6 to 7 months after initiation of therapy, neither case was fatal and no other etiology was found [60,61]. The most recent case suggested dose dependent causation, and a liver biopsy showed only mild portal lymphocytic inflammation and bile duct damage, which was compatible with drug hepatotoxicity [61]. Both cases of pioglitazone induced hepatic enzyme elevation resolved with cessation of the drug.

We must conclude that hepatotoxicity is unlikely to occur with either rosiglitazone or pioglitazone. Of the cases that have been reported involving rosiglitazone, only one suggested causation by the drug, but the patient was at an advanced age with other comorbidities and the two involving pioglitazone reversed with no apparent side-effects after discontinuation of the drug. A recent article found no hepatotoxic effects in 5508 clinical trial subjects who were taking rosiglitazone as monotherapy or combination [62].

Therefore, hepatotoxicity is not a class effect of the TZDs and seems to have been a unique side effect of troglitazone. Troglitazone’s chemical make up is different from the newer, safer formulations, and there are several, mainly speculative, theories as to troglitazone’s mechanism of causing hepatotoxicity. In vitro studies have shown that incubation of hepatic cells with troglitazone resulted in apoptotic cell death. This mechanism involved DNA fragmentation and nuclear condensation that was dependent on both the duration of exposure and tissue concentration of the drug [63]. Troglitazone also induces the cytochrome P450 isoform 3A4 and could be a predisposing factor to drug-drug interaction [64].

Hepatic dysfunction, acute hepatic necrosis, and hepatic death have been reported with other oral agents used to treat type 2 diabetes. Using the UKPDS database, an evaluation of the frequency of liver disease in DM2 patients as a result of oral therapy, mainly second-generation sulfonylureas and later metformin, was performed. Of the 44,406 DM2 patients only 57 were found to have a possibility of drug induced liver disease. Of these, 5l were continued on their oral agents without event and in only 2 patients were the oral hypoglycemic agents the only possible cause of liver disease [65]. However, there are case reports of both first and second-generation sulfonylureas causing hepatitis, cholestatic jaundice, hypersensitivity vasculitis and granulomatous hepatitis [66]. In particular, since 1980 three cases of granulomatous hepatitis secondary to glyburide have been reported [67,68]. Despite the large and long duration of exposure, only two deaths have been reported from glyburide liver toxicity [69]. The data on sulfonylurea induced hepatitis and hepatic necrosis is therefore underwhelming and should not affect the use of second-generation sulfonylureas, and monitoring of liver enzymes with these drugs is not warranted.

Acarbose is a poorly absorbed, reversible inhibitor of the small intestine brush-border alpha-glucosidases and has been reported to have hepatic toxicity. There have been nine occurrences of documented hepatotoxicity with most occurring in Spanish women [70]. All the cases demonstrate an idiosyncratic reaction with elevation of liver enzymes and bilirubin that resolve with cessation of therapy [71]. The reported cases did not include patients from the United States, and all but one case were in women between the ages of 40–65, five being Spanish, one being French, and three being Japanese [70–74]. Liver biopsy did not demonstrate any findings suggestive of an acute drug reaction and primary liver disease was ruled out with serology and imaging. The clustering among women and the ethnic predilection suggest the possibility of genetic predisposition. However, the increased number of regionally localized cases of hepatotoxicity could be explained by regional prescribing practices. Another possible cause of acarbose’s potential for hepatotoxicity is that whereas most of the ingested acarbose remains in the gut, a small amount is absorbed and excreted in the urine. When renal decompensation occurs and the absorbed acarbose excretion is decreased, serum and tissue levels of acarbose can elevate to levels that are toxic to the liver. The above case reports did not comment on the patients renal status so if we assume renal function was normal this would support an idiopathic drug reaction as the mechanism of the hepatotoxicity. Nevertheless, renal assessment should be performed before therapy and an alpha glucosidase inhibitor should not be used in patients with a serum creatinine exceeding 2.0 mg/dl. The Physicians Desk Reference 2002 recommends liver function tests every 3 months for one year and then “periodically.”

There is a commonly held misconception that metformin is hepatotoxic, but no case of metformin-induced hepatotoxicity has been reported in the more than 40 years that metformin has been available worldwide. However, metformin should not be utilized in those patients who have known liver disease or who are likely to develop acute hepatic toxicity (e.g. binge drinkers). The decreased liver function in these persons will prevent efficient metabolism of lactic acid and can result in lactic acidosis.


There is a significant association between diabetes mellitus, the liver, and liver disease. Hepatitis C infection appears to be a risk factor for the development of both diabetes and liver disease. Nonalcoholic hepatic steatosis is associated with the insulin resistance syndrome and can be partially reversed by lowering insulin resistance with medications or lifestyle interventions. Whether NASH (nonalcoholic steatohepatitis) can be reversed or its progression to cirrhosis arrested is under study at this time. Iron deposition from hemochromatosis may often precipitate diabetes and primary iron overload may be associated with insulin resistance and DM2. Because of involvement of the gallbladder with diabetic autonomic neuropathy there is an increased incidence of cholelithiasis, but a lower incidence of biliary colic and cholecystitis in the diabetic patient. Hepatotoxicity from medication used to treat diabetes is rare but more common when troglitazone was available.


1. Mehta SH, Brancati FL, Sulkowski MS, et al.: Letter: Prevalence of type 2 diabetes mellitus among persons with hepatitis C virus infection in the United States. Hepatology 2001; 33: 1554.
2. Williams I: Epidemiology of hepatitis C in the United States. Am J Med 1999; 107: 2S–9S.
3. Mehta SH, Brancati FL, Sulkowski MS, et al.: Prevalence of type 2 diabetes mellitus among persons with hepatitis C virus infection in the United States. Ann Intern Med 2000; 133: 592–9.
4. Ryu JK, Lee SB, Hong SJ, et al.: Association of chronic hepatitis C virus infection and diabetes in Korean patients. Korean J Intern Med 2001; 16: 18–23.
5. Ramji A, Yoshida, EM: Diabetes mellitus after liver transplantation: another extra hepatic manifestation of hepatits C. Can J Gastroenterol 2001; 15: 432.
6. Mason AL, Lau JYN, Hoang N, et al.: Association of diabetes mellitus and chronic hepatitis C virus infection. Hepatology 1999; 29: 328–333.
7. Sangiorgio L, Attardo T, Gangemi R, et al.: Increased frequency of HCV and HBV infection in type 2 diabetic patients. Diabetes Res Clin Pract 2000; 48: 147–51.
8. Caronia S, Taylor K, Pagliaro L, et al.: Further evidence for an association between non-insulin–dependent diabetes mellitus and chronic hepatitis C virus infection. Hepatology 1999; 30: 1059–63.
9. Allison MED, Wreghitt T, Palmer CR, et al.: Evidence for a link between hepatitis C virus infection and diabetes mellitus in a cirrhotic population. J Hepatol 1994; 21: 1135–1139.
10. Strassburg CP, Obermayer-Straaub P, Manns MP: Autoimmunity in hepatitis C and D virus infection. J Viral Hepatitis 1996; 3: 49–59.
11. McCullough, AJ: Update on nonalcoholic fatty liver disease. J Clin Gastroenterol 2002; 34: 255–62.
12. Marchesini G, Brizi M, Bianchi G, et al.: Association of Nonalcoholic Fatty Liver Disease with Insulin Resistance. American J Med 1999; 107: 450–455.
13. Dixon J, Bhathal PS, O’Brien PE: Nonalcoholic fatty liver disease: Predictors of nonalcoholic steatohepatitis and liver fibrosis in the severely obese. Gastroenterology 2001; 121: 91–100.
14. Reid AE: Nonalcoholic steatohepatitis. Gastroenterology 2001; 121: 710–23.
15. Bernal-Reyes R, Saenz-Labra A, Bernardo-Escudero R: [Prevalence of non-alcoholic steatohepatitis. Comparative study with diabetic patients]. Gastroenterol Mex 2000; 65: 58–62.
16. Valenti L, Fracanzani AL, Dongiovanni P et al.: Tumor necrosis factor alpha promoter polymorphism and insulin resistance in nonalcoholic fatty liver disease. Gastroenterology 2002; 122: 274–80.
17. Marchesini G, Brizi M, Bianchi G, et al.: Nonalcoholic fatty liver disease: a feature of the metabolic syndrome. Diabetes 2001; 50: 1844–50.
18. Ryysy L, Hakkinen AM, Goto T, et al.: Hepatic fat content and insulin action of free fatty acids and glucose metabolism rather thatn insulin absorption are associated with insulin requirements during insulin therapy in type 2 diabetic patients. Diabetes 2000; 49: 749–58.
19. Willner IR, Waters B, Patil SR, et al.: Ninety patients with nonalcoholic steatohepatitis: insulin resistance, familial tendency, and severity of disease. Am J Gastronenterol 2001; 96: 2957–61.
20. Dana SL, Hoener PA, Bilakovics JM, et al.: Peroxisome proliferatior-activated receptor subtype-specific regulation of hepatic and peripheral gene expression in the Zucker diabetic fatty rat. Metabolism 2001; 50: 963–71.
21. Cury D: EASD abstract #A271, Diabetologia 2000 (suppl).
22. Caldwell SH, Hespenheide EE, Redick JA, et al.: A pilot study of a thiazolidinedione, troglitazone, in nonalcoholic steatohepatitis. Am J Gastroenterol 2001; 96: 519–25.
23. Marchesini G, Brizi M, Bianchi G, et al.: Metformin in nonalcoholic steatohepatitis. Lancet 2001; 358: 893–4.
24. Ellervik C, Mandrup-Poulsen T, Nordestgaard BG, et al.: Prevalence of hereditary haemochromatosis in late-onset type 1 diabetes mellitus: a retrospective study. Lancet 2001; 358: 1405–9
25. Niederau C, Berger M, Stremmel W, et al.: Hyperinsulinaemia in non–cirrhotic haemochromatosis: impaired hepatic insulin degradation? Diabetologia 1984; 26: 441–4.
26. Turlin B, Mendler MH, Moirand R, et al.: Histologic features of the liver in insulin resistance-associated iron overload. A study of 139 patients. Am J Clin Pathol 2001; 116: 263–70.
27. Kwan T, Leber B, Ahuja S, et al.: Patients with type 2 diabetes have a high frequency of the C282Y mutation of the hemochromatosis gene. Clin Invest Med 1998; 21: 251–7.
28. Sampson MJ, Williams T, Heyburn PH, Green et al.: Prevelence of HFE (hemochromatosis gene) mutation in unselected male patients with type 2 diabetes. J Lab Clinc Med 2000; 135: 170–3.
29. Frayling T, Ellard S, Grove J, et al.: C282Y mutation in HFE (haemochromatosis) gene and type 2 diabetes. Lancet 1998; 351: 1933–4.
30. Moirand R, Mortaji AM, Loreal O, et al.: A new syndrome of liver iron overload with normal transferring saturation. Lancet 1997; 349: 95–7.
31. Czaja AJ, Carpenter HA, Santarch PJ, et al.: Genetic predispositions for the immunological features of chronic active hepatits. Hepatology 1993; 18: 816.
32. Homberg JC, Abuaf NA, Bernard O, et al.: Chronic active hepatitis associated with antiliver/kidney microsome antibody type 1: A second type of “autoimmune” hepatits. Hepatology 1987; 7; 1330–5.
33. Clemente MG, Meloni A, Obermayer-Straub P, et al.: Two Cytochromes P450 are major hepatocellular autoantigens in autoimmune polyglandular syndrome type 1. Gastroenterology 1998; 114: 324–8.
34. Blendis L, Brill S, Oren R: Commentary re: Hepatogenous Diabetes: Reduced insulin sensitivity and increased awareness. Gastroenterology 2000; 119: 1800–2.
35. Shetty A, Wilson S, Kuo P, et al.: Liver transplantation improves cirrhosis-associated impaired oral glucose tolerance. Transplantation 2000; 69: 2451–4.
36. Marchesini G, Bianchi GP, Forlani G, et al.: Insulin resistance is the main determinant of impaired glucose tolerance in patients with liver cirrhosis. Dig Dis Sci 1987; 32: 1118–24.
37. Proietto J, Dudley FJ, Aitken P, et al.: Hyperinsulinaemia and insulin resistance of cirrhosis: the importance of insulin hypersecretion. Clin Endocrinol (Oxf) 1984; 21: 657–65.
38. Sampelean D, Motocu M: Low insulin secretion in decompensated liver cirrhosis with diabetes mellitus. Rom J Intern Med 1993; 31: 265–9.
39. Kruszynska YT, Harry DS, Bergman RN, et al.: Insulin sensitivity, insulin secretion and glucose effectiveness in diabetic and nondiabetic cirrhotic patients. Diabetologia 1993; 36;121–8.
40. Petrides AS, Schylze-Berge D, Vogt C, et al.: Glucose resistance contributes to diabetes mellitus in cirrhosis. Hepatology 1999; 18: 284–91.
41. Petrides AS, Vogt C, Schulze-Berge D, et al.: Pathogenesis of glucose intolerance and diabetes mellitus in cirrhosis. Hepatology 1994; 19: 616–27.
42. Perseghin G, Mazzaferro V, Sereni LP, et al.: Contribution of reduced insulin sensitivity and secretion to the pathogenesis of hepatogenous diabetes: effect of liver transplantation. Hepatology 2000; 31: 694–703.
43. Merli M, Leonetti F, Riggio O, et al.: Glucose intolerance and insulin resistance in cirrhosis are normalized after liver transplantation. Hepatology 1999; 30: 649–54.
44. Poonawala A, Nair SP, Tuluvath PJ: Prevalence of obesity and diabetes in patients with cryptogenic cirrhosis: A case-control study. Hepatology; 32: 689–92.
45. Pazzi P, Scagliarini R, Gamberini S, et al.: Review Article: gall-bladder motor function in diabetes mellitus. Aliment Pharmacol Ther 2000; 14(Suppl 2): 62–5.
46. Chapman BA, Chapman TM, Frampton CCM, et al.: Gallbladder volume: Comparison of diabetic and controls. Dig Dis Sci 1998; 43: 344–8.
47. De Santis A, Attilit AF, Ginanni Corradini S, et al.: Gallstones and diabetes: A case-control study in a free-living population sample. Hepatology 1997; 25: 787–90.
48. Meguro T, Shimosegawa T, Satoh A, et al.: Gallbladder emptying and cholecystokinin and pancreatic polypeptide responses to a liquid meal in patients with diabetes mellitus. J Gastroenterol 1997; 3295: 628–34.
49. Mitsukawa T, Takermura J, Ohgo S, et al.: Gallbladder function and plasma cholecystokinin levels in diabetes mellitus. Am J Gastroenterol 1990; 85: 981–5.
50. Bedirli A, Sozuer EM, Yuksel O, et al.: Laproscopic cholecystectomy for symptomatic gallstones in diabetic patients. J Laparoendosx Adv Surg Tech 2001; 11: 281–4.
51. Aucott JN, Cooper GS, Bloom AD, et al.: Management of gallstones in diabetic patients. Arch Intern Med 1993; 153: 1053–8.
52. Zein NN, Abdulkarim AS, Wiesner RH, et al.: Prevalence of diabetes mellitus in patients with end-stage liver cirrhosis due to hepatitis C, alcohol, or cholestatic disease. J Hepatol 2000; 32: 209–17.
53. Adami HO, Chow WH, Nyren O, et al.: Excess risk for primary liver cancer in patients with diabetes mellitus. J Natl Cancer Inst 1996; 88: 1472–7.
54. El-Serag HB, Richardson PA, Everhart JE: The role of diabetes in hepatocellular carcinoma: A case-control study among United States Veterans. Am J Gastroenterology 2001; 96: 2462–8.
55. Graham DJ, Drinkard CR, Shatin D: Liver enzyme monitoring in patients treated with troglitazone, JAMA 2001; 286: 831–2.
56. Nierenberg DW: “Did this drug cause my patient’s hepatitis?” and related questions, Ann Int Med 2002; 136: 480–3.
57. Gouda HE, Khan A, Schwartz J, et al.: Liver failure in a patient treated with long-term rosiglitazone therapy. Am J Med 2001; 111: 584–5.
58. Forman LM, Simmons DA, Diamond RH: Hepatic failure in a patient taking rosiglitazone. Ann Intern Med 2000; 132: 118–21.
59. Al-Salman J, Arjomand H, Kemp DG, et al.: Hepatocellular injury in a patient receiving rosiglitazone. A case report. Ann Intern Med 2000; 132: 121–4. Errata - Al-Salman, et al, Correction: Liver Injury and Rosiglitazone. Ann Intern Med 2000; 133: 237
60. Maeda K: Letter: Hepatocellular injury in a patient receiving Pioglitazone. Ann Intern Med 2001; 135: 306.
61. May LD, Lefkowitch JH, Kram MT, et al.: Mixed hepatocellular-cholestatic liver injury after pioglitazone therapy. Ann Intern Med 2002; 136: 449–52.
62. Lebovitz HE, Kreider M, Freed MI: Evaluation of Liver Function in Type 2 Diabetic Patients During Clinical Trials. Diabetes Care 2002; 25: 815–21.
63. Yamamoto Y, Nakajima M, Yamazaki H, et al.: Cytotoxicity and apoptosis produced by troglitazone in human hepatoma cells. Life Sci 2001; 70: 471–82.
64. Sheen AJ: Hepatotoxicity with thiazolidinediones: is it a class effect? Drug Saf 2001; 24: 873–88.
65. Jick SS: Frequency of liver disease in type 2 diabetic patients treated with oral antidiabetic agents. Diabetes Care 1999; 22: 2067–71.
66. Rumboldt Z, Bota B: Favorable effect of glibenclamide in a patient exhibiting idiosyncratic hepatotoxic reactions to both chlorpropamide and tolbutamide. Acta Diabetol Lat 1984; 21: 387–91.
67. Saw D, Pitman E, Maung M, et al.: Granulomatous hepatitis associated with glyburide. Dig Dis Sci 1996; 41: 322–5.
68. Ingelmo M, Vivancos J, Bruguera M, et al.: [Hypersensitivity vasculitis and granulomatous hepatitis induced by glybenclamide: a case report]. Med Clin (Barc) 1980; 75: 306–8.
69. Van Basten JP, van Hoek B, Zeijen R, et al.: Glyburide-induced cholestatic hepatitis and liver failure. Case-report and review of the literature. Neth J Med 1992; 40: 305–7.
70. Andrade RJ, Lucena MI, Rodriguez-Mendizabal M: Letter: Hepatic injury caused by acarbose. Ann Inter Med 1996; 124: 931.
71. Andrade RJ, Lucena M, Vega JL, et al.: Letter: Acarbose-associated heptotoxicity. Diabetes Care 1998; 21: 2029.
72. Fujimoto Y, Ohhira M, Miyokawa N, et al.: Acarbose-induced hepatic injury. Lancet 1998; 351: 340.
73. Carrascosa M, Pascual F, Aresti S: Acarbose-induced acute severe hepatotoxicity. Lancet 1997; 349: 698–9.
74. Mannecier D, Zafrani ES, Dhumeaux D, et al.: [Acarbose-indusce acute hepatitis]. Gastroenterol Clin Biol 1999; 23: 1398–9.
© 2003 Lippincott Williams & Wilkins, Inc.