Thiazolidinedione therapy for nonalcoholic steatohepatitis: Go, stop, or proceed with caution?§


  • Stephen A. Harrison M.D.

    Corresponding author
    1. Division of Gastroenterology and Hepatology, Brooke Army Medical Center, Fort Sam Houston, TX
    • Brooke Army Medical Center, Department of Gastroenterology, 3851 Roger Brooke Drive, Fort Sam Houston, TX 78234===

    Search for more papers by this author
    • fax: 210-916-5611

  • See Article on Page 445

  • Disclaimer statement: The opinion of ascertains contained herein are the private views of the authors and are not to be construed as official or reflecting the view of the Department of the Army or the Department of Defense.

  • §

    Potential conflict of interest: Nothing to report.

The epidemic of obesity has been associated with a significant increase in nonalcoholic fatty liver disease (NAFLD). The pathogenesis of NAFLD in this setting is predicated on the premise that obesity-related insulin resistance is responsible for the development of hepatic steatosis. Our current understanding holds that in some patients, the increased free fatty acid (FFA) flux to the liver and decreased hepatic FFA oxidation results in lipotoxicity and progression to hepatocyte ballooning, lobular inflammation, and pericellular fibrosis—the histopathologic hallmarks of nonalcoholic steatohepatitis (NASH). To this end, investigators have predominantly focused on therapies that improve insulin resistance.


FFA, free fatty acid; LDL, low-density lipoprotein; HDL, high-density lipoprotein; NAFLD, nonalcoholic fatty liver disease; NAS, NAFLD activity score; NASH, nonalcoholic steatohepatitis; PPAR-γ, peroxisome proliferator-activated receptor-γ; TZD, thiazolidinedione.

Weight loss has been recommended for many years, and there is data to show that this therapy is efficacious. Bariatric surgery improves the underlying metabolic dysfunction seen in the morbidly obese patient and improves histopathology in most studies.1 In others, a modest weight loss (∼5%) improves insulin resistance while a weight loss of ∼10% is associated with improvement in steatosis, ballooning, inflammation, and NAFLD activity score (NAS).2 Unfortunately, the majority of patients with NAFLD are unable to lose weight and maintain their weight loss. Consequently, therapies aimed at improving insulin resistance either through augmenting or supplanting weight loss have been studied. The thiazolidinedione (TZD) class of insulin sensitizers has been the focus of attention for the past few years.

Rosiglitazone and pioglitazone, both TZDs, were approved in 1999 for the treatment of type II diabetes. They are peroxisome proliferator-activated receptor-γ (PPAR-γ) agonists. PPAR-γ receptors are located predominantly in adipose tissue, but can also be found elsewhere, to include pancreatic β cells, vascular endothelium, and to a lesser extent in liver and skeletal muscle.3 The TZD mechanism of action is not completely understood, but they improve insulin resistance in liver, adipose tissue, and muscle. Data suggest that the TZDs decrease FFA flux to the liver and improve visceral adiposity in part through an increase in subcutaneous adipose tissue mass and up-regulation of specific adipocytokines such as adiponectin.3 Adiponectin expression, decreased in the setting of obesity, type II diabetes, metabolic syndrome, cardiovascular disease,4 and NAFLD,5 is increased by PPAR-γ agonists resulting in reduced hepatic gluconeogenesis as well as improved hepatic fatty acid oxidation (via increased adenosine monophosphate–activated protein kinase) and increased glucose disposal in skeletal muscle.4 Adiponectin also reduces inflammation, in part, by blocking nuclear factor-κB and inhibiting the release of proinflammatory cytokines6 and may suppress hepatic stellate cell proliferation.7

Recent evidence suggests that there are also differences between the two TZDs, at least when it comes to lipid metabolism.8 Pioglitazone has been shown to decrease plasma triglycerides, increase high-density lipoprotein (HDL), reduce low-density lipoprotein (LDL) concentration, and increase LDL particle size8 and decrease hepatic de novo lipogenesis by up to 40%.9 Rosiglitazone, alternatively, has no effect on hepatic de novo lipogenesis9 and actually has been shown to raise plasma LDL concentration and does not reduce triglyceride concentrations.8 This may explain, at least in part, why pioglitazone has positive cardiovascular effects (improved carotid intimal medial thickness10 and coronary atheroma volume11) whereas rosiglitazone does not. Thus, it appears that the two TZDs are not completely similar and pioglitazone may have additional PPAR-α activity.3

Focusing on NAFLD, both TZDs have now been studied in prospective, placebo-controlled trials ranging from 6-24 months duration. The results have been varied (Table 1). The TZDs most reliably improve steatosis, just as they do with insulin resistance. However, there are variable responses to other histopathologic lesions commonly seen in NASH to include lobular inflammation, ballooning degeneration, and fibrosis. Intuitively, we would all like to see fibrosis regression. However, no study has conclusively shown TZD therapy to improve fibrosis. Alternatively, no study has shown significant progression of fibrosis either, albeit the follow-up periods are relatively short. Improvement in ballooning degeneration with the TZDs has been varied. Rosiglitazone improved ballooning in a small uncontrolled study,12 but in a larger controlled trial, no improvement in ballooning was found.13 Similarly, two relatively small, placebo-controlled trials with pioglitazone of 6-12 months duration using 30 mg daily in one trial and 45 mg daily in another, found different results.5, 14 The lower dose pioglitazone study did not show significant improvement within the pioglitazone group,14 but the higher dose study found improved ballooning within and between the two groups.5 Improvement in inflammation among the randomized controlled trials was only shown with pioglitazone.5, 15 This finding may be of particular importance because a recent study found that inflammation was an independent predictor of fibrosis progression.16 Interestingly, a larger placebo-controlled trial with pioglitazone in which patients were treated for 24 months using the 30 mg dose showed improvement in steatosis and inflammation but not in ballooning (P = 0.08).15 Prior to the publication by Ratziu et al.17, no studies of greater than 12 months duration existed for rosiglitazone.

Table 1. Randomized, Placebo-Controlled Trials
StudynDesignDrugDoseDurationInsulin SensitivityImproved Histopathology
  • *

    Ballooning did not improve statistically in the pioglitazone arm, but due to worsening of ballooning in the placebo arm, the difference between the two arms was significant.

Belfort et al.555RCTPioglitazone45 mg6 monthsImprovedSteatosis, inflammation, ballooning, NAS, fibrosis (P = 0.08)
Aithal et al.1474RCTPioglitazone30 mg12 monthsNot improvedBallooning*
Sanyal et al.15 (PIVENS)247RCTPioglitazone30 mg24 monthsImprovedSteatosis, inflammation, (ballooning, P = 0.08)
Ratziu et al.1363RCTRosglitazone8 mg12 monthsImprovedSteatosis

In the current issue of HEPATOLOGY, Ratziu et al report on 40 patients who underwent liver biopsies a minimum of 2 years after starting rosiglitazone therapy for NASH.17 The patients were part of a previous trial in which rosiglitazone was compared prospectively to placebo for the treatment of NASH. All patients originally received study drug for 1 year. In the current trial, 22 of the original placebo patients were then put on roglitazone and 18 of the original patients on rosiglitazone were continued on the drug. All patients were then followed for 2 years and repeat biopsies were obtained. Overall, insulin sensitivity, as defined by the Homeostasis Model Assessment (HOMA) score, improved by 30% and serum alanine aminotransferase improved by 24% over the study period. Among the original placebo patients, steatosis decreased by a median of 15% (P < 0.001) over the 2-year follow-up period, but no improvement in necroinflammation or fibrosis was seen. Among the patients who continued on rosiglitazone for 2 additional years (3 years total), no further improvement in steatosis (20% improvement originally) was found, nor was there improvement in the NAS or fibrosis. No significant changes in serum triglycerides or HDL cholesterol were seen.

With the new data that we now have on the TZDs and NASH, it is worthwhile to stop and ponder whether we are moving in the right direction. Will the TZDs change the natural history of NASH as they appear to do with diabetes and possibly coronary artery disease? Both TZDs have shown a decrease in progression rates to diabetes in those with gestational diabetes or impaired fasting glucose/impaired glucose tolerance.18, 19 In addition to the trials demonstrating positive cardiovascular effects with pioglitazone,10, 11 a large meta-analysis of more than 16,000 diabetic patients has shown a significant (18%) reduction in death, myocardial infarction, or stroke when treated with pioglitazone.20 NAFLD is linked to the development of diabetes and coronary artery disease.21, 22 In fact, heart disease appears to be the leading cause of death among patients with NAFLD.23 Thus, even if TZD therapy does not result in significant quantifiable histopathologic improvement in NASH, it is possible that there could be delay in progression to diabetes or symptomatic coronary artery disease.

As shown in the current study, the histopathologic improvement seen with rosiglitazone appears to reach its maximum benefit within the first 12 months of therapy. Lengthening therapy beyond this point did not result in further improvement. This data is consistent with a recent 5-year prospective study in bariatric surgery patients which showed that the greatest improvements in steatosis and ballooning occurred within the first year of surgery.24 Could we use this data in designing future clinical trials for NASH? If significant improvement is defined as a two-point improvement in the NAS, then a 12-month study endpoint may be appropriate. If fibrosis improvement is the goal, then a longer study duration may be required.

The current study highlights another important question among patients with NASH who are treated with TZDs. Why is there such heterogeneity in the histopathologic response seen among the studies done to date? There are several potential explanations. There may be inherent differences in the histopathologic response between the two TZDs, just as there are in lipoprotein metabolism. Additionally, the dose of pioglitazone and the treatment duration are different among the three prospective, randomized, placebo-controlled trials evaluating this drug. Pioglitazone has demonstrated a dose-response curve in relation to its glucose-lowering/insulin-sensitizing effects that may also apply to its histologic benefit. Furthermore, the patient populations are varied in relation to ethnicity, gender prevalence, age, geography, and diabetes prevalence. These factors may also contribute to the varied histopathologic responses.

Characterization of the patients who respond or do not respond to TZD therapy is lacking. This study by Ratziu et al.17 begins to shed some light in this area. They found that patients who improved their insulin sensitivity by 40% or greater had a significant improvement in hepatocyte ballooning. Alternatively, those with less of a response did not do as well, corroborating the experience seen in the bariatric surgery field.24 These findings highlight the particular need to better characterize patients who may or may not respond to TZD therapy, similar to how we characterize patients with hepatitis C prior to initiating therapy with pegylated interferon and ribavirin. Does gender, ethnicity, or the presence or absence of diabetes mellitus among patients with NASH influence the degree of insulin sensitivity improvement seen with the TZDs? Does the degree of baseline insulin resistance influence histopathologic response rates? Is there more to the story than simply improving insulin resistance? The TZDs clearly do not improve histology in every patient. With further knowledge of response factors, clinicians may be able to distinguish or even predict responders from nonresponders prior to initiating therapy with a TZD or within a few weeks to months of starting therapy depending on the insulin response curve.

Finally, the results from the PIVENS trial presented at the recent 60th annual meeting of the American Association for the Study of Liver Diseases suggest that treatment strategies other than insulin sensitivity improvement may also improve histopathology.15 The multiple mechanisms leading to NASH may require that more than one therapeutic approach be considered in the treatment of this disease. The future may involve either targeted monotherapy based on specific patient demographics and baseline laboratory data or combination therapy that is aimed at improving key pathogenic abnormalities seen in patients with NASH, such as oxidative stress, insulin resistance, apoptosis, or other, as-yet clarified, manifestations of lipotoxicity.

For now, lifestyle intervention remains the cornerstone of therapy for NAFLD. However, for those patients who are unable to exercise or lose weight through diet, or for those patients with more advanced NASH, studies with adjuvant pharmacologic therapy targeting insulin resistance appears indicated. More recent data suggest that patients with NAFLD are not just at risk for progression to cirrhosis and hepatocellular carcinoma, but also are at significant risk for progression to diabetes and cardiovascular disease. Given this information, it may be myopic for Hepatologists to simply focus on the histopathologic effects of NASH therapies. The beneficial effects of the TZDs to prevent or delay the progression to diabetes and the potential beneficial effects of pioglitazone on cardiovascular disease should be considered concurrently with the potential for histopathologic improvement. Consequently, although further studies are warranted examining the role of TZDs, particularly focused on histological dose-responses, identifying response factors, and length of therapy, we should be cautiously optimistic for the future of TZDs (pioglitazone >> rosiglitazone) in the treatment of NAFLD. Long-term risks of bone fractures, peripheral weight gain, and the potential to develop congestive heart failure remain with both TZDs. Therefore, the risk-benefit ratio should be considered in each patient prior to initiating adjuvant therapy with a TZD.