Non-alcoholic fatty liver disease (NAFLD) is associated with obesity and insulin resistance (IR) and encompasses a spectrum of disease. Hepatic steatosis alone is usually benign but the combination of steatosis, hepatocyte injury and inflammation, which comprises non-alcoholic steatohepatitis (NASH), can cause progressive fibrosis and cirrhosis. Although all patients with NAFLD have increased cardiovascular mortality, only patients with portal fibrosis are at increased risk of liver-related mortality. Fibrosis in NASH is believed to be mediated by lipotoxicity via excess oxidative stress driven by excess fatty acid (FA) traffic in the setting of obesity and IR. Oxidative stress also induces replicative arrest and senescence, the latter being characterised by a senescence-associated secretory phenotype (SASP) that may contribute to disease progression via the secretion of pro-inflammatory factors.
Although hepatocyte injury in NASH is parenchymal and early fibrosis in adults is typically centrilobular, progressive NASH is characterised by portal fibrosis, accompanied by an as yet uncharacterised chronic inflammatory infiltrate, and in paediatric patients pure portal fibrosis is the classical lesion. The reasons for these findings are unknown. In acute injury states, hepatocytes divide to restore parenchymal mass but an increasing burden of senescence limits this compensatory mechanism. In chronic injury, parenchyma is replaced via the proliferation of a periportal compartment of hepatic progenitor cells (HPCs) that can differentiate into mature hepatocytes or cholangiocytes. This is associated with a ductular reaction (DR), which represents a proliferation of small keratin-7 (K7) positive biliary ductules and a transit-amplifying population of HPCs. Although typical of cholestatic diseases, a DR is also seen in parenchymal diseases (e.g. Hepatitis C, NASH). Furthermore, HPC expansion and a DR are associated with the presence of portal fibrosis, suggesting they contribute to disease progression.
This thesis aimed to explore several areas of uncertainty and determine the importance of key concepts in portal fibrogenesis by: first, determining whether simple clinical criteria can be used to distinguish patients with portal fibrosis; second, by assessing histological associations with different patterns of NASH fibrosis in adults and children; third, by developing an in vitro model of hepatocyte senescence to characterise the hepatocyte SASP and determine the effects of secreted factors on other cell types; and fourth, by characterising the portal inflammatory infiltrate in NASH to determine the type of immune response associated with portal fibrosis. The most significant results are summarised below:
(i) Four unweighted clinical criteria (“Easy Score”, comprising age>40, body mass index>28, evidence of IR, and aspartate aminotransferase:alanine aminotransferase (AST:ALT) ratio>0.8) were applied to a test cohort (n=53) and validation cohort (n=227) of patients with biopsy-proven NAFLD. An Easy Score≥ 2 identified portal fibrosis in these cohorts with sensitivities of 100.0% and 92.0%, respectively. Performance of the Easy Score was comparable to other non-invasive scoring models. Apart from age > 40, all parameters of the Easy Score, as well as ALT > 40 and platelets < 180, remained significantly associated with the presence of portal fibrosis after stepwise multiple logistic regression.
(ii) Histological features of paediatric (n=38) and adult (n=56) NAFLD were assessed. The DR and HPC numbers were scored using K7 immunostaining in each case, and post-treatment (rosiglitazone or gastric banding) biopsies were also assessed in all adults. Heterogeneity of fibrosis patterns was prominent in children, and in some adults (n=10) a previously unrecognised pattern of diffuse subsinusoidal fibrosis (DSSF) was apparent, usually after gastric banding (9 of 10 adults, P<0.001). In children, portal fibrosis was associated with more extensive DR (P=0.021) and higher HPC numbers (P<0.001), whereas centrilobular fibrosis was associated with lobular inflammation (P=0.026) and hepatocyte ballooning (P=0.001).
(iii) An in vitro model of hepatocyte senescence was developed by treating HepG2 cells with sub-lethal doses of hydrogen peroxide (H2O2; 300 μM daily for 2 days). Seven days after release from treatment, treated cells demonstrated increased heterochromatin 1-beta (HP1β; P<0.001)) and p21 (P<0.001) compared with untreated HepG2 cells, consistent with senescence. Excess FAs (1 mM for 48 hours) also induced senescence in HepG2 cells.
(iv) Using the H2O2-treated HepG2 cell model in comparison with untreated HepG2 cells, senescent hepatocytes were found to secrete increased levels of protein (12.8 ± 5.8 fold increase; P=0.03). Preliminary characterisation of the hepatocyte SASP was undertaken via microarray of extracted RNA, as well as examination of secreted proteins using one-dimensional (1D) and 2-dimensional (2D) gel electrophoresis of hepatocyte conditioned media (CM), quantification of protein expression using stable isotope labelling with amino acids in culture (SILAC), and analysis of exosome-derived proteins. Amongst up-regulated proteins, apolipoprotein E demonstrated consistently increased expression over multiple modalities. Senescent hepatocyte CM (HPCM) significantly induced the migration of THP1-derived M1 and M2 macrophages (P<0.001 and P<0.05, respectively). Exposure of M1 macrophages to HPCM led to a reduction in expression of pro-inflammatory genes, including IL-10 (P=0.028), IL-6 (P<0.0001), IL-10 (P<0.01) and TNFα (P<0.001), suggesting a “switch” to the
more pro-fibrogenic M2 phenotype.
(v) The portal inflammatory infiltrate was characterised by immunostaining adult NAFLD liver sections (n=40) for CD4 and CD8 (T cells), CD20 (B cells), CD56 (NK cells), CD68 (macrophages) as well as IL-17+ cells. Replicative arrest (p21) and the DR (K7) were also assessed. Two mouse models of fatty liver were also used to study associations between the inflammatory infiltrate, the DR and portal fibrosis and the time course of portal changes. The portal inflammatory infiltrate in human NAFLD comprised predominantly CD8+ T cells and macrophages. Lobular macrophage numbers correlated with total p21 staining (rs=0.450, P=0.024) suggesting macrophages may provide a nexus between parenchymal senescence and portal fibrogenesis. Absolute numbers of T cells (CD4+ and CD8+), B cells and macrophages increased significantly with stage of fibrosis (all P<0.05), although patients with portal fibrosis demonstrated a significant proportionate increase in CD4+ T cells only (P<0.05). Significant correlations between the grade of DR and stage of fibrosis (rs=0.756; P<0.0001) or grade of portal inflammation (rs=0.472; P<0.01) were seen in humans, and the DR appeared to precede fibrosis in the murine model. Significant increases in portal IL-17+ cells were seen in association with increasing stage of fibrosis (P=0.018) and increased Th-17-associated gene expression was seen in association with the presence of portal fibrosis, including IL-6 (P<0.001), TGFβ (P<0.05), TNFα (P<0.05) and IL-17 (P<0.01), suggesting the inflammatory response comprises a Th-17 response.
Taken together, these findings provide evidence for a number of mechanisms that may contribute to disease progression in NAFLD. The likelihood of multiple fibrogenic pathways in NASH and the inter-related roles of senescence, the DR/HPC expansion, and the portal inflammatory response in NASH warrant further investigation.