Nonalcoholic fatty liver disease (NAFLD) is the most prevalent liver disease in this country and a major global health problem that is likely to worsen due to the epidemic of obesity and diabetes.1 No established therapy exists for NAFLD, but one experimental approach has been to treat with agents that increase insulin sensitivity. This strategy is based on the concept that the mechanism underlying the development of NAFLD and other components of the metabolic syndrome is insulin resistance.1 Prominent among the insulin-sensitizing agents being examined as NAFLD treatments has been the thiazolidinedione class of antidiabetic drugs which includes pioglitazone. A randomized, placebo-controlled short-term study of pioglitazone in nondiabetic patients with NAFLD recently demonstrated significant metabolic and histologic improvement with a reduction in liver injury and fibrosis in the pioglitazone-treated patients.2
The thiazolidinediones are ligands for the nuclear receptor peroxisome proliferator-activated receptor γ (PPARγ) which mediates the effects of these drugs.3 PPARγ is a transcription factor that regulates critical cellular processes such as metabolism, proliferation, differentiation, and inflammation. Thiazolidinediones decrease insulin resistance, because PPARγ activation increases insulin sensitivity through a complex series of effects involving multiple organs but centered primarily on the adipocyte. The highest cellular levels of PPARγ are in adipocytes in which activation of this nuclear receptor induces white adipocyte differentiation and lipid storage. Increased sequestration of lipid in adipose tissue lowers serum free fatty acid levels,4 decreasing lipid delivery to other organs such as the liver and skeletal muscle where deleterious lipid deposition may occur and promote insulin resistance. PPARγ activation also up-regulates adipocyte production of adiponectin which can act to increase hepatic glucose uptake and suppress glucose production.5, 6 Finally, stimulation of the PPARγ pathway inhibits adipocyte production of tumor necrosis factor-α (TNFα) and resistin, which mediate insulin resistance.7
Although the levels are much lower than in adipose tissue, PPARγ is expressed in normal liver, and levels increase in rodent models of steatosis. Gavrilova et al.8 delineated the relative contributions of adipocyte versus hepatocyte PPARγ effects in response to the thiazolidinedione rosiglitazone by examining its differential effects in wild-type and lipoatrophic mice with normal or a knockout of hepatocyte PPARγ expression. In the absence of fat tissue (lipoatrophic mice), rosiglitazone induced hepatic steatosis that was PPARγ-mediated because fat accumulation failed to occur in mice with a hepatocyte-specific PPARγ knockout. In mice with normal fat tissue, the effect of rosiglitazone was to decrease hepatic lipid content through a mechanism independent of hepatocyte PPARγ because an equivalent effect occurred in hepatocyte PPARγ knockout mice. Thus, the effect of isolated PPARγ activation in the liver was to promote lipid accumulation, but the effect of PPARγ on fat tissue superseded the direct hepatic effects and led to a decrease in hepatic steatosis.
The macrophage is another cell in which PPARγ activation may mediate the therapeutic effects of thiazolidinediones in NAFLD.9 PPARγ inhibits macrophage activation and proinflammatory cytokine production,9 and therefore acts in both adipocytes and macrophages to decrease the levels of cytokines that mediate peripheral and hepatic insulin resistance and liver inflammation in NAFLD. The critical function of macrophage PPARγ signaling in the maintenance of insulin sensitivity is evident because a macrophage-specific knockout of PPARγ is sufficient to induce insulin resistance and hepatic steatosis.10 Thus, PPARγ-dependent effects of thiazolidinediones on macrophages as well as adipocytes likely mediate the potential benefits these drugs have on insulin responsiveness and hepatic lipid accumulation.
In this issue of HEPATOLOGY, novel studies by Aoyama and colleagues11 examine the physiological effects of pioglitazone on partial hepatectomy–induced liver regeneration in the KK-Ay mouse model of NAFLD. Partial hepatectomy is associated with transient hepatic lipid accumulation that has been proposed to provide the energy required for cell proliferation.12 Controversy exists on whether hepatic steatosis impairs regenerative potential. Obese mice with a genetic loss of leptin signaling have a defective regenerative response to partial hepatectomy,13 although the inability of exogenous leptin to reverse this defect despite resolution of the hepatic steatosis suggests that increased hepatic lipid content may not be the mechanism of this impairment.14 Both normal15 and defective16 hepatic regeneration has been demonstrated in mice with diet-induced NAFLD. Finally, recent studies by Newberry et al.17 have suggested that regeneration after partial hepatectomy is unaffected by hepatic lipid content, because mouse models with a wide range of hepatic triglyceride levels failed to exhibit any differences in proliferation.
The studies by Aoyama et al.11 were conducted in mutant KK-Ay mice, which develop obesity, insulin resistance, and hyperleptinemia with leptin resistance in the absence of a genetic defect in leptin signaling. In theory, the mice therefore represent a good rodent model of human NAFLD. These mice exhibited a profound regenerative defect, because for 48 hours after partial hepatectomy they had absolutely no increase in hepatocyte bromodeoxyuridine staining or cyclin D1 expression. There was also significant mortality in the KK-Ay mice which was attributed to impaired regeneration. It seems unlikely, however, that failed regeneration was the only mechanism as some mice died within 12-24 hours, well before regeneration occurs normally.
Pioglitazone pretreatment augmented the hepatic proliferative response in KK-Ay mice in response to partial hepatectomy. Bromodeoxyuridine and proliferating cell nuclear antigen staining was markedly increased with pioglitazone treatment, albeit to levels still well below those in wild-type mice. This effect was achieved with a short 5-day course of pioglitazone. Mortality from partial hepatectomy was also blocked by pioglitazone. Interestingly, after partial hepatectomy, the previously hyperglycemic KK-Ay mice developed hypoglycemia and marked hyperinsulinemia that was blocked by pioglitazone treatment. The paradoxical decrease in serum glucose in the diabetic animals along with the sudden increase in insulin production in response to partial hepatectomy is difficult to explain. These findings do suggest that the prosurvival effect of pioglitazone may have been mediated by its metabolic effects that stabilized insulin responsiveness and gluconeogenesis. The marked increase in adiponectin levels in the pioglitazone-treated animals may have mediated this effect, as well as having a direct effect on hepatocellular proliferation.18 The hypoglycemia in the previously hyperglycemic mutant mice suggests that they might have become acutely more insulin sensitive, and hepatic insulin hypersensitivity has been reported in diabetic rats after partial hepatectomy.19 Energy homeostasis may have been severely disrupted in the liver and other organs in KK-Ay mice after partial hepatectomy, which could have led to both impaired regeneration and death. To address this possibility, it would be interesting to examine hepatic adenosine triphosphate levels in wild-type and untreated and pioglitazone-treated animals as well as the effects of glucose supplementation on mortality and the regenerative defect in mutant mice.
In keeping with the known ability of PPARγ to inhibit inflammatory responses, pioglitazone treatment blunted increases in TNFα and interleukin-6 (IL-6) after partial hepatectomy. These two cytokines have beneficial proliferative and hepatoprotective effects, so it seems counterintuitive that pioglitazone-induced decreases in these factors were beneficial. However, the mutant mice had abnormal levels of these cytokines after partial hepatectomy with a markedly increased early and a second later peak in TNFα messenger RNA expression, and a normal early rise but an additional second peak in serum IL-6. Hyperproduction of these cytokines may have been deleterious, and although the excess TNFα did not lead to apoptotic liver injury as indicated by the absence of cytokeratin-18 cleavage, TNFα may have inhibited proliferation. In addition, a more careful assessment for liver injury, including that resulting in necrosis, should be performed. KK-Ay mice had a normal early induction of signal transducer and activator of transcription 3 (STAT3) phosphorylation, but STAT3 phosphorylation was sustained, an effect that was prevented by pioglitazone. Depending on the hepatic cell type undergoing STAT3 activation, prolonged STAT3 signaling may have had beneficial or detrimental hepatic effects.20 Assuming that pioglitazone's effects were PPARγ-mediated, studies should be performed in selective cellular PPARγ knockouts to determine whether the drug's proliferative effects are mediated by hepatocytes, nonparenchymal cells, and/or extrahepatic cells.
Of note is that prior studies in partially hepatectomized mice have demonstrated that thiazolidinediones, including pioglitazone, decreased regeneration.21, 22 However, in contrast to the investigations by Aoyama et al., the studies were conducted in normal animals. The current finding of the opposite effect in a fatty liver further suggests that steatosis may alter the regenerative response. The data available from these studies are not sufficient to suggest a mechanism for this differential effect. If direct hepatic effects of pioglitazone are involved, one simple explanation may be differences in receptor number. PPARγ levels are low in normal rodent and human liver and decrease further in response to partial hepatectomy.22 In contrast, obese and diabetic animals, including KK-Ay mice, have increased PPARγ levels. More detailed studies contrasting the signaling responses of normal versus steatotic animals to PPARγ activation may provide a better understanding of whether proliferative responses are altered in fat-laden hepatocytes.
The novel findings in the study by Aoyama et al. add to the potential beneficial effects of pioglitazone on the steatotic liver (Fig. 1). It is impressive that significant metabolic and regenerative improvements were achieved with a short course of therapy. Whether these effects can translate to humans, in whom PPAR signaling often differs from rodents, remains to be determined. First, the effects of pioglitazone after partial hepatectomy or on a more modest proliferative response need to be examined in other rodent NAFLD models. However, while awaiting further experimental evidence that pioglitazone may in fact promote hepatocyte proliferation in the setting of a fatty liver, it is interesting to speculate how a proliferative effect of pioglitazone may expand its use. Pioglitazone may particularly benefit older individuals with NAFLD in whom liver regenerative capacity may also be impaired by aging. Patients with steatotic livers without frank steatohepatitis may benefit from a preoperative course of pioglitazone before undergoing resective hepatic surgery. Steatotic donor livers for transplant might be preconditioned with pioglitazone to reduce primary nonfunction that may result in part from impaired regeneration.23 Although these therapeutic uses are highly speculative and their value must be weighed against potential side effects, the fact that thiazolidinediones may have additional beneficial effects on a steatotic liver beyond insulin sensitization does provide further hope that these agents will prove efficacious in the treatment of human NAFLD.