Not all models of fatty liver are created equal: Understanding mechanisms of steatosis development is important

Medical research is often constrained by the limited availability of both healthy and diseased tissue, making animal models an essential tool for studies of disease pathogenesis. Ideally, animal models should not only reproduce the established pathological expression of the human disease, but also recapitulate the context within which a disease develops and progresses. These issues are pertinent to hepatology research, in particular the study of non-alcoholic fatty liver disease (NAFLD) and its consequences. Many animal models of NAFLD are available, but the majority develop steatosis only, with few progressing to the inflammatory condition of steatohepatitis.1 In this issue of the Journal, Veteläinen and colleagues describe the differences in NAFLD development in two superficially similar dietary models: rats fed choline-deficient (CD) or methionine (referred to as methione in the article by Veteläinen et al.) and choline-deficient (MCD) diets.2 CD diet–fed rats develop steatosis in association with weight gain and impaired insulin sensitivity, whereas MCD-fed rats go on to develop steatohepatitis associated with weight loss and oxidative stress. MCD feeding appears to cause steatosis by inhibiting hepatic lipid export. In combination, methionine and choline deficiency block the hepatic synthesis of phosphatidylcholine, an essential phospholipid for very low-density lipoprotein (VLDL) synthesis and secretion.3 In contrast, choline deficiency alone has recently been shown to not significantly inhibit hepatic lipid export,4 and therefore reduced VLDL synthesis and secretion is unlikely to be the cause of steatosis in this model. The study by Veteläinen and colleagues highlights the importance of understanding the mechanisms responsible for steatosis development within animal models, and comments on the implications of this for hepatology researchers. For example, the use of leptin or leptin receptor–deficient models of steatosis for investigating the effects of hepatic lipid accumulation on liver regeneration is inappropriate, as leptin itself plays a direct role in these processes.5 Similarly, for ischaemia-reperfusion injury, the use of the MCD model needs to be carefully assessed as oxidant stress plays a key role in steatohepatitis development in this model.6, 7 As oxidant stress is also a key pathogenic pathway in ischaemia-reperfusion injury, the MCD model may compound injury due to oxidant stress, rather than from hepatic lipid accumulation alone. Further, as discussed by Veteläinen et al., models of NAFLD that develop uncomplicated steatosis may underestimate the adverse effects that steatohepatitis may have on hepatic injury, regeneration, and liver fibrosis. The comparison of CD and MCD-fed rodents by Veteläinen et al. demonstrates the well-known need for suitable animal models of NAFLD, that encompass the whole spectrum of disease from steatosis to progressive, fibrosing steatohepatitis. Although the MCD model is a popular model for studying inflammatory recruitment, cell injury, and fibrosis in steatohepatitis, it is not a model of metabolic syndrome,8 which is now generally accepted as the key determinant of nonalcoholic steatohepatitis.9, 10 It is therefore of limited use for determining the metabolic defects in hepatic lipid partitioning that initiate NAFLD. In recent years, a number of novel animal models of steatohepatitis have been described, within which benign steatosis and steatohepatitis characterized by lobular inflammation, steatosis and ballooning degeneration of hepatocytes, develop under varying dietary circumstances. One such model is the intragastric overnutrition model in which an implanted gastrostomy tube delivers caloric overload.11 All over-fed mice develop metabolic syndrome and steatosis, with a subset progressing to steatohepatitis; this scenario is reminiscent of over-nutrition in obese humans.9 Another promising model is the foz/foz mouse, which develops obesity, metabolic syndrome, and steatosis when fed standard rodent chow, but progresses to steatohepatitis when administered a diet high in saturated fats.12 This model is also reflective of human NAFLD, as metabolic syndrome is present and serum adiponectin levels are reduced in obese mice with steatosis, and further reduced in mice with steatohepatitis. Interestingly, oxidant stress does not appear to be linked to steatohepatitis development or progression in either the gastric over-nutrition or foz/foz models of NAFLD,11, 12 in contrast to observations in MCD-fed rodents.13 Thus, the view of oxidant stress as the ‘second hit’ of steatohepatitis development may not always be correct,14 at least for these models. As obesity and NAFLD become increasingly prevalent in Asia and elsewhere, it is important to understand not only the mechanisms of steatosis and steatohepatitis development, but also the complications they introduce to other liver conditions, such as hepatitis C. Animal models can play a useful role in the endeavor to answer these questions, but it is of critical importance to select models that replicate causative disease mechanisms as well as disease pathology and clinical outcomes. The ideal animal model will encompass the full spectrum of fatty liver disease, in the context within which NAFLD develops in humans: obesity, metabolic syndrome, and associated hypoadiponectinaemia.15 Veteläinen and colleagues have shown the limitations of two commonly used models. Choline deficiency leads to obesity and insulin resistance (which could be partly related to the high sucrose content of the diets), and steatosis also develops, but there is no transition to steatohepatitis. Instead, insidious fibrosis leads to cirrhosis.16 In contrast, methionine and choline deficiency does lead to steatohepatitis with progressive hepatic fibrosis, but animals lose weight and have increased insulin sensitivity rather than insulin resistance and metabolic syndrome. Thus, while these models have taught us much about the pathogenesis of steatosis and of inflammatory recruitment in one experimental context, it is time that the focus of experimental approaches to understand NAFLD/NASH pathogenesis moves to more appropriate models of this disease in humans.