Background
It has recently been estimated that human non-alcoholic fatty liver disease (NAFLD) has a 25% worldwide prevalence and is expected to become the major reason for liver transplantations in the western world [
1,
2].
NAFLD ranges from simple steatosis to non-alcoholic steatohepatitis (NASH), the latter characterized by additional inflammation, hepatocellular hydropic degeneration (also called ballooning) and eventually Mallory–Denk bodies associated with fibrosis which can lead to cirrhosis. Type 2 diabetes is considered the primary risk factor for progression of simple steatosis to advanced stages of NASH [
3]. In addition, NAFLD seems to be a central risk factor for the development of cardiovascular disease [
4] and other complications also related to diabetes [
5,
6]. It can be difficult to diagnose NAFLD given that clinical signs usually are sparse even at late stages where liver damage is substantial; and changes in circulating liver biomarkers are often non-specific. Histopathological evaluation of samples from liver biopsies is still considered the gold standard for diagnosing the disease even though new non-invasive alternatives are emerging [
7].
The hepatic disease mechanisms are still largely unknown, and effective treatment modalities are lacking. Therefore, various animal models have been investigated to help unravel the pathogenesis and for testing new pharmaceutical drug candidates or lifestyle interventions. Diet- or chemically-induced rodent models of NASH/NAFLD have mainly been used, and several mouse strains exist with spontaneous or transgenic mutations that pinpoint different signalling pathways [
8,
9]. Larger animal models have also been examined, as anatomy, physiology or simply size can be beneficial depending on the research question of interest [
10,
11].
Histopathological features resembling NASH in humans have been reported in Ossabaw minipigs [
10,
12] fed a diet high in fat, fructose and cholesterol. Others have studied Bama minipigs and Microminipigs on high fat/high sucrose diet or high fat/high cholesterol diet with a supplement of cholic acid, respectively, and reported microvesicular steatosis and inflammation, but none or limited hepatocellular ballooning and fibrosis [
11,
13]. The Göttingen Minipig is widely used in studies of obesity and metabolic syndrome [
14‐
16], but diet-induced histopathological changes in the liver have not been characterized.
In this study, the aim was to investigate the liver changes in Göttingen Minipigs fed a diet with high content of fat, fructose and cholesterol for over a year. Hepatic gross morphology, histopathology and tissue content of lipids and glycogen as well as changes in relevant circulating biomarkers were investigated. Furthermore, it was evaluated if diabetes would exacerbate the alterations and if a diet change to standard diet would limit pathologic changes. The hypothesis was that Göttingen Minipigs fed a diet high in fat, fructose and cholesterol would develop hepatic changes that resemble human NAFLD including NASH.
Discussion
This study demonstrated that Göttingen Minipigs fed a high fat diet with fructose and cholesterol became obese and developed hepatomegaly with hepatic fibrosis, inflammation and cytoplasmic alterations when compared to animals fed a normal diet. However, only few animals developed marked steatosis, and hepatocellular hydropic degeneration (ballooning) and Mallory–Denk bodies were not observed in any animal. Streptozotocin-induced diabetes did not exacerbate the changes in circulating biomarkers or hepatic histopathology, compared to non-diabetic animals fed a similar diet. The group changed to standard diet for 6 months had no hepatic changes, except from excess collagen deposition.
The large number of animals, the more than 1-year study period, and the comprehensive investigation of several parameters in the present study strengthen the characterization of the hepatic changes in obese Göttingen Minipigs with and without diabetes.
As expected, the FFC and FFC
DIA groups developed obesity with higher BW, BF% and dyslipidemia (defined as increased plasma TG and TC) compared to the two groups fed standard diet. The FFC
DIA group had significantly higher levels of fasting GLU and FRA confirming hyperglycemia. Plasma ALP and GLDH were increased, whereas an increase in circulating concentrations of ALT and AST was not found, in fact plasma ALT was lower in FFC pigs compared to SD and FFC/SD. Studies of Ossabaw minipigs on the same diet also found no difference in ALT compared to control animals, whereas AST was increased. However, in pigs, GLHD is considered a more reliable marker of acute liver damage, as compared to ALT, which is non-specific in pigs [
26]. Also, in human patients with NAFLD, ALT values have not been found to correlate with the degree of histopathological changes [
27]. Table
7 provides an overview of relevant NAFLD/NASH characteristics including selected circulating biomarkers from human and pig studies.
Table 7
Comparison of selected elements of porcine models of NAFLD/NASH and human NAFLD/NASH
Steatosis, macro | Yes | Yes | Minimal | 3 out of 7 | No | No | – |
Steatosis, micro | No | No | No | Yes | – | Yesa | Yes |
Fibrosis, perisinusoidal | Yes/No | No | Yes | Yes | Yes | – | No |
Fibrosis, portal | No | Yes | Yes | Yes | Yes | Yes | No |
Inflammation, lobular | Yes | No | Yes | 4 out of 7 | No | Yes | Yes |
Inflammation, portal | No | Yes | Yes | No | – | Yes | – |
Hepatocellular ballooning | Yes | No | No | Yesb | Yesb | No | Yesb |
↑ triglycerides liver content | – | – | No | Yes | No | – | Yes |
↑ cholesterol liver content | – | – | Yes | – | Yes | – | Yesc |
↑ ALT | Yes/No | Yes/No | No | No | No | Yes | No |
↑ AST | – | Yes/No | No | Yes | Yes | Yes | Yes |
Hypercholesterolemia | – | – | Yes | Yes | Yes | Yes | Yes |
Hypertriglyceridemia | Yes | Yes | Yes | Yes | Yes1 | Yes | No |
The LW was highly increased in both FFC and FFCDIA groups compared to SD, and despite elevated liver content of cholesterol in both groups, unexpectedly no statistically significant differences were found for triglycerides liver content among the four groups.
Furthermore, the degree of histopathological alterations was less than expected, especially regarding steatosis, despite the presence of metabolic disturbances indicated by the changes in circulating markers. Few animals displayed more than 5% macrovesicular steatosis which is a criterion for the diagnosis of NAFLD in humans [
28,
29]. Rodent models on a high fat diet usually report abundant macro- and microvesicular steatosis [
30,
31], but previous studies in Ossabaw minipigs on the same diet (5B4L) also reported lack of macrovesicular steatosis, despite extensive liver injury [
10,
12]. Others have reported extensive microvesicular steatosis with little or no presence of macrovesicular steatosis in pigs on high fat plus high cholesterol or high sucrose diets [
11,
13]. Microvesicular steatosis was rarely seen in our pigs, and the low level of both types of steatosis is consistent both with the biochemical analysis and the quantification of lipid on ORO stained slides. This modest accumulation of lipids in the liver could reflect the fact that the liver in pigs, in contrast to humans and rodents, is not the primary site of de novo lipogenesis [
32]. Also, the overnight fasting period before termination might have had some influence on the lipid content in the liver. However, so far it is unclear what is responsible for the significantly increased LW in FFC and FFC
DIA groups and it needs to be clarified if hepatocyte hypertrophy or hyperplasia is the reason for the present hepatomegaly.
Hepatocellular ballooning, another criterion for the diagnosis of human NASH, was not seen in our study, but extensive cytoplasmic alterations in hepatocytes were present in FFC and FFC
DIA groups. It was speculated that these alterations could be caused by an accumulation of glycogen. Glycogenic hepatopathy is a condition with massive cytoplasmic deposition of glycogen in hepatocytes leading to hepatomegaly, and is mostly seen in patients with poorly controlled type 1-diabetes. The diagnosis is confirmed with liver biopsies and staining for glycogen with PAS [
33]. In this study, staining for glycogen with PAS however did not show correlation with CA in either FFC
DIA or FFC, and the FFC
DIA group also showed decreased glycogen content in their liver tissue in comparison to the FFC group. Instead a strong association between CA and cholesterol content in the liver persisted, and perhaps CA could reflect a functional adaptation to the increased cholesterol load. In humans, it has recently been suggested that dietary cholesterol activates the hepatic stellate cells thereby promoting fibrosis especially if hepatocyte uptake or biliary excretion of cholesterol is inhibited [
34].
Mallory–Denk bodies is a characteristic feature of NASH in humans, but can be difficult to identify and often additional IHC staining has to be performed, e.g., using ubiquitin or cytokeratin 8/18 antibody. Lackner et al. found that CK 8/18 was diminished or absent in ballooned hepatocytes compared to normal hepatocytes. They also found that Mallory–Denk bodies were not always present [
25]. Others suggest double immunohistochemical staining of CK8/18 and ubiquitin for the optimal detection of hepatocellular injury in human liver [
35]. In some mouse models of NASH, Mallory–Denk bodies have been reported morphologically [
30,
36,
37], but until now no previous studies with pigs [
10‐
12] or other rodent species [
31,
38] have identified Mallory–Denk bodies along with hepatocellular ballooning. Immunohistochemical staining of CK8/18 or ubiquitin in liver tissue of porcine NAFLD/NASH models has not been reported. However, in a mouse model of NAFLD/NASH staining for CK8/18 has been used [
31] and found less or absent staining in hepatocytes with ballooning.
The predominantly periportal fibrosis seen in our minipigs is coherent with findings in Ossabaw minipigs [
12]. This is in contrast to humans where perisinusoidal fibrosis around the central vein dominates and periportal fibrosis occurs only in later disease stages. A possible explanation for this peripheral deposition of collagen could be the localization of the porcine hepatic stellate cells. These cells are responsible for collagen production when activated and heterogeneity of stellate cells in the porcine liver was described by Wake et al. [
39] who found that desmin positive stellate cells were most abundant in the peripheral regions of the classical lobules. Interestingly, two subtypes of NAFLD have been described in pediatric patients; type 1 resembles ‘adult’ NAFLD whereas type 2 is characterized by steatosis, portal inflammation and portal fibrosis [
40] and is known to be the most prevalent type in children. The mechanism behind this pediatric portal fibrosis is unknown but perhaps a common pathway for portal fibrosis in children and pigs exist. The hepatic inflammation seen in FFC and FFC
DIA groups was not a reflection of a systemic inflammatory state as the circulating inflammatory markers CRP and ALB showed no difference between the four groups.
Interestingly, the FFC/SD group also differed from SD in terms of fibrosis, despite being fed the same standard diet for the last 6 months. This indicates that fibrosis developed in the first 7 months of high fat/high cholesterol-feeding and either did not progress further during the 6 months of healthy dieting, or maybe even regressed.
Diabetes did not exacerbate the hepatic histopathology, although the FFCDIA pigs had substantial dyslipidemia with elevated levels of TC and TG and a liver tissue cholesterol content exceeding that of the FFC group. Previous unpublished studies indicated that 2% cholesterol was needed in order to achieve the desired elevated level of circulating total cholesterol in normal minipigs, whereas diabetic minipigs reached the same level on a 1% cholesterol diet, which was also the case here. Six diabetic pigs were terminated prematurely nearly reducing the group by half, which could have led to lack of statistical power when comparing the FFCDIA group with the FFC group.
Our diabetic animals had a type 1-like diabetes phenotype and were treated with long acting insulin analogue to keep blood glucose around 15 mM. A lower incidence of NAFLD has been found in patients with type 1 diabetes compared to patients with type 2 diabetes, and it is suggested that insulin treatment’s inhibiting effect on lipolysis is responsible for the decreased level of free fatty acids accumulating in the liver. Insulin resistance induced by excessive caloric intake is also known to play a role in hepatic lipid accumulation and the development of NAFLD. Only FFC
DIA had elevated fasting blood glucose, but insulin resistance can be present in muscle and liver without hyperglycemia in human patients with NAFLD [
41]. Although insulin resistance was not directly assessed in the current study, K
G, AUC
insulin and HOMA-IR values indicated that FFC and likely FFC
DIA animals were insulin resistant on a whole-body level, but since no profound steatosis was present, hepatic insulin resistance is less likely to be a major factor in this model.
A limitation to this study was, that 11 animals ended the study prematurely and necropsy reports mentioned enlarged pale livers in all FFC (n = 2) and FFCDIA (n = 6) pigs, but no further analyses were performed. It could be speculated that these excluded animals especially from the FFCDIA group may have exhibited more pronounced hepatic changes thus biasing our results as only animals with less pronounced changes were able to complete the full study period. Another limitation is that no liver biopsies were taken at different time points during the study, especially before the FFC/SD group was changed to standard diet, making it difficult to conclude if the differences between FFC and FFC/SD were due to lack of progression or regression of marked changes already present at the intervention time point.
Overall, this diet-induced obese Göttingen Minipig model with or without diabetes poses some challenges in terms of translatability with human NASH, because some of the cardinal characteristics (abundant steatosis, hepatocellular ballooning and Mallory–Denk bodies, and zone 3 fibrosis) were missing. However, to explain some of the rather unexpected elements of this model, further studies are needed. Additional staining with e.g. CK8/18, ubiquitin, p62 or sonic hedgehog markers could be performed in order to validate if Mallory–Denk bodies (or their precursors) were present in this minipig model. Moreover, evaluation of serum CK8/18 may elucidate if depletion at tissue level could give as consequence an increase in the circulation of keratin fragments that are major components of Mallory–Denk bodies [
42]. By elucidating the pathological pathways perhaps combined with gene expression and electron microscopy, the mechanisms leading to CA in hepatocytes and the portal/periportal fibrosis could be further clarified.
Authors’ contributions
CSP: planning and performing experiment, data analyses including histopathological assessment and imaging analyses, interpretation, writing manuscript. BØC: study design, planning experiment, interpretation, writing manuscript. RKKI: planning and supervising histopathological assessment and imaging analyses, interpretation, writing manuscript. TPL: study design, planning and performing experiment, data analyses, interpretation, writing manuscript. NEZ: interpretation, writing manuscript. HDP: study design, planning experiment, interpretation, writing manuscript. MV: planning and supervising histopathological assessment, interpretation of histopathological data, writing of manuscript. LHO: study design, planning of experiment, supervising data analyses, interpretation, writing manuscript. All authors read and approved the final manuscript.