Environmental stressors such as obesity and insulin resistance, exposure to hepatotoxins, and hepatitis C virus infection increase hepatocyte lipid content, with a significant change in triglyceride, cholesterol as well as other lipid species. Resultant fatty liver disease (FLD) variably progresses to steatohepatitis, cirrhosis and hepatocellular carcinoma (HCC).1 Evidence for shared determinants of FLD and end-stage liver complications first arose from human genetics,2 showing that the main risk variants for FLD had a proportional impact on fibrosis and HCC, and fibrosis/HCC development was very closely linked to the impact on the lipid content of hepatocytes.3, 4 However, this theory is challenged by the observation that hepatic fat accumulation and biochemical markers of liver damage and inflammation tend to decrease with the progression of liver fibrosis.5-7 Interpretation of these observational studies is challenging, as they tend to be cross-sectional in nature and often include participants that already have cirrhosis.8

In patients affected by FLD, hepatic fat content is progressively reduced (even to within the normal range) in individuals with the most advanced liver disease,9 defining the concept of ‘burnt-out nonalcoholic steatohepatitis’. Indeed, demonstration of the presence of excess hepatic fat is no longer a requirement to diagnose metabolic dysfunction associated with fatty liver disease in individuals with cirrhosis, whereas in the past these patients were considered to be affected by ‘cryptogenic’ disease.10 Several explanations have been proposed to explain the loss of hepatic fat with fibrosis progression and development of portal hypertension, including alterations in hepatic vasculature and mitochondrial metabolism, reduced exposure to insulin and stimulation of catabolic pathways,7 and increased levels of adiponectin, the insulin-sensitizing adipokine which is cleared by the liver.11 Conversely, despite the progressive reduction in hepatic fat, severe fibrosis is consistently linked to insulin resistance and to a higher incidence of diabetes.5, 12, 13

In a seminal paper recently published in Nature, Ng et al. reported new findings that may provide an explanation for this paradox.14 The study aimed at the identification of somatic mutations occurring in non-malignant liver tissues (n=29) from individuals with advanced alcoholic and nonalcoholic FLD, which were extensively sampled (vs controls with a normal liver (n=5). The key finding was the detection of convergent somatic mutations of genes involved in the regulation of lipid metabolism, namely FOXO1, CIDEB and GPAM, in the liver of patients with advanced FLD.14 The predicted impact of these genetic variants is to shift substrates away from lipid storage. These data suggest that these mutations might have been clonally selected as advantageous during chronic lipotoxic exposure in hepatocytes. For example, the authors observed putative loss-of-function mutations in GPAM which encodes the acyl-CoA:glycerol-sn-3-phosphate acyltransferase 1 enzyme (GPAM/GPAT1), an enzyme that catalyses the initial and rate-limiting step of triglyceride synthesis in situations of positive energy balance. Consistently, Gpam loss of function in mice protects from the development of fatty liver.15 In keeping with a prominent role of GPAM in the regulation of hepatic lipid metabolism, another GPAM missense variant, rs2792751 (p.Val43Ala), has recently been associated with increased alanine transaminase, hepatic steatosis and upregulation of lipid metabolism pathways, and included among the common inherited determinants of fatty liver disease in Europeans.16 GPAM deficiency thus would be predicted to result in a reduction in the liver fat content, independent of the progression of liver disease, and a compensatory carbon flux to gluconeogenesis and apparent higher insulin resistance.

Another intriguing finding was that the most frequent and recurrent genetic alterations were located at a hotspot in forkhead transcription factor O1 (FOXO1), a mediator of insulin signalling at transcriptional level, through AKT-dependent phosphorylation and nuclear exclusion.17 In fact, a single FOXO1 mutation (p.Ser22Trp or S22W) was highly overrepresented in several individuals and in multiple independent clones derived from the same liver, suggesting genetic convergence. FOXO1 Ser22 has been shown to be phosphorylated by AMPK, which subsequently prevents AKT-mediated FOXO1 phosphorylation and degradation18; thus, this mutation would be predicted to be loss-of-function. On the contrary, Ng et al. presented data in hepatoma cell lines demonstrating that the S22W substitution impaired insulin-dependent nuclear export, leading to constitutive retention of active FOXO1 in the nucleus by abolishing consensus sequence for the 14-3-3 nuclear export protein.14 A likely reconciliation of these two findings is that the bulky tryptophan group acts similarly to AMPK-mediated phosphorylation, both preventing FOXO1/14-3-3 binding.

Nuclear trapping of FOXO1 has already been implicated in the increased gluconeogenesis seen with caloric excess and oxidative stress.19 Similarly, FOXO1 is upregulated in human NASH, in parallel with oxidative stress and insulin resistance.20 Upregulation of antioxidant responses may therefore account for the selective advantage of acquiring the constitutively nuclear FOXO1 bearing the S22W variant.17, 21 In addition, in experimental models, FOXO1 shifts hepatocytes metabolism from an anabolic state (glucose uptake, lipid synthesis and lipid droplet storage) to a catabolic state (lipid remodelling, lipolysis and stimulation of gluconeogenesis and glucose release22-26). These data are in line with metabolomic data from hepatoma cells overexpressing S22W, which show increased expression of transcriptional targets and of gluconeogenesis.14 Thus, increased glucose production and hepatic insulin resistance may represent a detrimental by-product of the loss of hepatic fat.27

Limitations of the Ng et al. work include (a) the low number of individuals and the limited spectrum of disease surveyed, precluding generalisation of the results; (b) the lack of understanding on how the S22W FOXO1 mutation arises in relation to specific drivers of liver disease; and, most importantly, (c) demonstration of a causal relationship between the FOXO1 S22W mutation and the decrease in hepatic fat typical of fibrosis progression, and evaluation of the impact on pathways underlying disease progression.28

Nonetheless, these findings have several important potential implications. First, burnt-out NASH may be a natural response to chronic lipotoxicity. In addition, the decrease in hepatic fat with FLD progression may at least be partially mediated by genetic mechanisms, that is somatic mutagenesis. In fact, these data represent the first evidence that somatic mutagenesis, an increasingly recognized mechanism of non-neoplastic disease,29 shapes the biology of chronic degenerative liver conditions distinct from cancer. The potential role of somatic mutagenesis in the decrease of hepatic fat during liver disease progression is depicted in Figure 1. Although additional mechanistic and clinical studies are warranted to test the hypothesis that FOXO1 S22W and other convergent somatic mutations underpin burnt-out steatohepatitis, and to examine the prognostic and therapeutic implications, these findings validate a key and foreground role of FOXO1 and other metabolic factors in liver disease.

Hypothetical mechanism linking somatic mutagenesis to decrease hepatic fat content during fatty liver progression to severe fibrosis

Italian Ministry of Health (Ministerodella Salute), RicercaFinalizzata RF-2016-02364358 (‘Impact of whole exome sequencing on the clinical management of patients with advanced nonalcoholic fatty liver and cryptogenic liver disease’), (LV). Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Ricerca corrente LV, DP, ALF). Fondazione IRCCS Ca’ Granda core COVID-19 Biobank (RC100017A), ‘Liver BIBLE’ (PR-0391) (LV). Innovative Medicines Initiative 2 joint undertaking of European Union’s Horizon 2020 research and innovation programme and EFPIA European Union (EU) Programme Horizon 2020 (under grant agreement No. 777377) for the project LITMUS (LV). The European Union, programme ‘Photonics’ under grant agreement ‘101016726’ (LV). Gilead_IN-IT-989-5790 (LV).