Review Article

Shared Disease Mechanisms Between Nonalcoholic Fatty Liver Disease and Metabolic Syndrome

Translating Knowledge From Systems Biology to the Bedside

Silvia Sookoian; Carlos J. Pirola


Aliment Pharmacol Ther. 2019;49(5):516-527. 

In This Article


The pathogenesis of complex diseases, including NAFLD and Metabolic syndrome, is explained by perturbations of several molecular processes that originate from the interplay between environmental factors and genetic susceptibility.

Recent technological advances, including OMICs approaches for the exploration of quantitative molecular traits such as gene, protein and metabolite expression, and genome-wide characterisation of DNA sequence variation, have dramatically changed the understanding of human diseases not only in terms of underlying mechanisms but also in terms of diagnosis and treatment. This is in part due to, for example, increasing ability to identify molecular mediators associated with a given trait.

In this work, through systems biology modelling, we provided evidence that NAFLD and Metabolic syndrome share disease mechanisms, including a network of molecular targets. The analysis of reactome pathways of the shared molecular mediators shows fairly straightforward results that are consistent with modulation of immune response and SUMOylation of intracellular receptors. While involvement of immune-related processes in the development of the Metabolic Syndrome is not necessarily a novel finding, post-translational modifications of nuclear receptors, like SUMOylation of proteins, represent an interesting and poorly explored mechanism to explain the pathogenesis and clinical course of NASH, as well as its interaction with related comorbidities. For instance, there is evidence indicating that SUMOylation of nuclear receptors, such as LXRα and LXRβ, affects not only inflammatory pathways, tissue injury and repair[20,21] but also mediates resistance to therapies targeting them.[22]

A remarkable finding based on the reactome pathways analysis is the overrepresentation of platelet-related processes, such as platelet degranulation, activation, signalling and aggregation. Platelets are key effectors of vascular damage[23] and their role in NASH seems to be relevant as well. For instance, we have demonstrated that NASH is associated with changes in platelet abundance of TGFβ1 mRNA.[12] In our cohort of patients with NAFLD and Metabolic syndrome, circulating platelet TGFB1 mRNA levels were also significantly associated with insulin resistance.[12] Platelet-released PDGFa/b may be also relevant in hypoxia response, which is important in the development of hepatocarcinoma.[24]

Gene-based disease-enrichment analysis confirmed the presence of shared underlying molecular effectors for major systemic disorders, not only diseases of the Metabolic Syndrome but also autoimmune diseases, kidney, respiratory and nervous system disorders, cancer, and infectious diseases. There are interesting examples of this concept in the literature that are worth highlighting. For example, the association of PNPLA3-rs738409 with the susceptibility to diverse liver-related phenotypes has been extensively replicated.[25] Nevertheless, it would appear that the rs738409 variant may also exhibit pleiotropic effects. Recently published results of a large phenome-wide association study (PheWAS) have revealed that rs738409 modestly increases the risk of type 2 diabetes (Odds ratio 1.08).[16] Conversely, it also appears that rs738409 has a protective effect against acne (Odds ratio 0.90), gout (Odds ratio 0.92) and gallstones (Odds ratio 0.95), and was shown to be associated with decreased levels of total cholesterol (Odds ratio 0.96) and LDL cholesterol (Odds ratio 0.97).[16] While the biological relevance of the reported odds ratios—which indicate marginal effects on the highlighted phenotypes—remains to be determined, it seems quite interesting to prove these findings in diverse cohorts across the world.

In addition, the identified list of shared genes showed different degrees of genetic overlap with inherited diseases/disorders/or traits annotated in the OMIN database, including laboratory abnormalities and neoplasias. OMIN also contains annotations of genes associated with the development of late-onset inherited diseases, among which there are hundreds of disorders; some examples of late-onset inherited diseases are Alzheimer's, Parkinson's and Huntington's disease, and even some Mendelian forms of the common diseases, such as the Metabolic syndrome components. Indeed, a remarkable observation is that, in the pairwise comparison between loci associated with NAFLD and those associated with Metabolic syndrome-related traits, there were significant differences in the proportion of the list of genes potentially implicated in late-onset disorders. It seems that obesity-associated loci comprise the highest proportion (17.7%) of genetic associations with late-onset inherited diseases compared to NAFLD (9.6%), followed by dyslipidemia (14.4%), type 2 diabetes (7.5%) and arterial hypertension, which presents the lowest proportion (5.4%).

These observations have important clinical implications, as the triad of obesity, type 2 diabetes and NAFLD accumulates ∼40% of loci associated with susceptibility of late-onset diseases. In fact, obesity across lifespan is associated with increased risk of brain atrophy and late-onset Alzheimer's disease.[26] Likewise, glycosylated haemoglobin levels above 6.5% increase the risk of incident Alzheimer's disease by 2.8-fold,[26] and NAFLD is associated with a smaller total cerebral brain volume.[27]

We also identified a gene-drug interaction network associated with the list of genes/proteins shared among NAFLD, inflammation, fibrosis and Metabolic syndrome. The predicted interactions, which are based on drugs with already known mechanisms of action and validated structure and pharmacological properties, may provide useful insights for drug repositioning, as well as for the implementation of combined therapeutic approaches.

Likewise, our gene-drug interaction network could support the rationale for selecting a single drug with wide range of biological effects, for example statins that in addition of its effects in reducing serum cholesterol levels present pleiotropic systemic effects, including modulation of proinflammatory cytokines, reactive oxygen species and platelets reactivity[28] as well as effects on the liver, including antifibrotic properties.[29] Systemic[30] and liver-related[31,32,33] pleiotropic effects can be equally observed by the blockade of the renin-angiotensin-aldosterone system. Nevertheless, the complexity of our predicted reactome pathways shared among NAFLD, inflammation, fibrosis and Metabolic syndrome suggests that placing much emphasis on a particular drug or molecular target might not be the route to therapeutic success unless a "magic bullet" is discovered.

However, the concepts above discussed–shared pathways for disease as opposed to therapeutic approaches (single vs multi-drug treatment of NASH)—are not mutually exclusive. In fact, the multiplicity and complexity of the involved processes highlight that summative and or synergistic therapeutic interventions can result in beneficial effects.[34]

Finally, there were remarkable examples of genes and/or proteins in the list of 50 shared loci—for example, APOA1, haptoglobin and SOD2 (Superoxide Dismutase 2)—that are highly expressed in liver, heart and pancreas; therefore, they seem to be relevant to organ physiology. Deregulation of gene and protein expression is highly important for understanding the pathogenesis of diseases. One may argue that our systems biology strategy could be biased towards genes or proteins associated with abnormal expression, as terms used for searching biomedical databases were essentially those that refer to diseases. Nevertheless, by showing that genes/proteins in the shared list have widespread expression in different tissues under physiological conditions, we may assume that our approach is not affected by reverse causality (diseases expression).

Our strategy has some limitations that are inherent to the systems biology strategy, as it relies on data that have been annotated or deposited in databases. Furthermore, it could be argued that MEDLINE and ENTREZ records relate to heterogeneous knowledge that includes human studies and experimental results based on cell lines or animal models. Some loci might have also been omitted due to inconsistencies between gene identifiers and/or gene symbols. However, it is important to highlight that these potential shortcomings are ameliorated in part by the enrichment analysis.

In conclusion, we show that clinical traits are linked by molecular networks. The high degree of pleiotropy in the list of shared genes/proteins supports the hypothesis that clinical phenotypes (diseases) should be replaced by a broader concept of pathophenotypes. Most importantly, we leveraged on the shared pathogenic mechanisms among NAFLD and the Metabolic Syndrome diseases for providing evidence that NAFLD, specifically NASH, requires multi-target therapeutic approaches, rather than focusing on single mechanism/s of disease.