Article Text

Download PDFPDF

Non-alcoholic fatty liver disease as a risk factor for hepatocellular carcinoma: mechanisms and implications
  1. Felix Stickel1,
  2. Claus Hellerbrand2
  1. 1Department of Visceral Surgery and Medicine, Inselspital, University of Bern, Bern, Switzerland
  2. 2Department of Internal Medicine I, University of Regensburg, Regensburg, Germany
  1. Correspondence to Professor Claus Hellerbrand, Department of Medicine I, University Hospital Regensburg, Franz-Josef-Strauß-Allee 11, D-93053 Regensburg, Germany; claus.hellerbrand{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The pathophysiological significance of hepatic lipid accumulation in the absence of significant alcohol consumption is increasingly recognised. Thus, non-alcoholic fatty liver disease (NAFLD) is now considered the most common cause of liver enzyme elevation in Western countries.1 It is regarded as the hepatic manifestation of the metabolic syndrome (MS), characterised by central obesity and insulin resistance (IR), and resulting diabetes type 2, dyslipidaemia and hypertension.2 NAFLD encompasses mild hepatic steatosis to steatohepatitis (non-alcoholic steatohepatitis (NASH)) with significant necroinflammation and progressive fibrosis. NAFLD is believed to account for a large fraction, if not entirely for what was previously termed ‘cryptogenic cirrhosis’.3

Since cirrhosis is the main risk factor for hepatocellular carcinoma (HCC), liver cancer could be simply a complication of end-stage NAFLD, similar to the situation encountered in other chronic fibrosing liver diseases. However, accumulating evidence suggests that hepatocarcinogenesis may also be related to earlier stages of NAFLD. Case series of patients with HCC and NAFLD as the only identified risk factor strongly suggest that hepatocarcinogenesis is part of the natural history of NAFLD. Based on the known association of NAFLD with IR and MS, approximately two-thirds of the patients were obese and/or diabetic,4 and a remarkable 25% of these patients had no cirrhosis. Considering the rapidly increasing prevalence of both conditions in affluent societies, and their significance in the pathophysiology of NAFLD, a rising incidence of NAFLD and its complications—including HCC—can be expected in the mid-term future. Therefore, it is particularly worrying that the most persuasive evidence for an association between NAFLD and HCC derives from studies on the risk of HCC in patients with MS. In a large prospective cohort study HCC mortality was significantly higher in obese subjects than in those with normal body mass index.5 The most comprehensive data underlining the significance of diabetes as an important risk factors for HCC originate from a large Veteran Affairs study, which showed that the risk of HCC doubled in patients with diabetes compared with those without.6 Interestingly, the risk of developing HCC was particularly high in individuals with long-term diabetes preceding the diagnosis of HCC. The promoting effect of diabetes on the evolution of HCC has since been confirmed in other studies recently reviewed in a meta-analysis, which suggests that diabetes increases HCC risk before the development of cirrhosis.7 Further, in two studies in which patients with known liver disease/cirrhosis had been excluded, diabetes was confirmed as a risk factor for HCC.6 8 Also experimental NAFLD models demonstrate that HCC may occur without cirrhosis. In genetically obese mice, liver hyperplasia occurs at the earliest stages of NAFLD, and with ageing these mice develop an increased incidence of HCC in the absence of overt hepatic inflammation and fibrosis.9 Moreover, a clear progression from NASH, to cirrhosis and cancer has been described in another rodent model.10

The close association between NAFLD and the MS, on the one hand, and obesity and diabetes with HCC, on the other, raises the question whether it is ‘truly’ NAFLD or rather the associated MS that promotes hepatocarcinogenesis. However, it is extremely plausible that hepatic steatosis itself ‘lubricates’ HCC development, although we are just beginning to dissect the underlying mechanisms.

Mechanisms linking NAFLD with hepatocarcinogenesis

Lipid accumulation in hepatocytes and hepatoma cells induces cancer-related molecular signalling involving nuclear factor kappaB (NF-κB) and c-Jun N-terminal kinase(JNK)/activator protein-1 activity, and overexpression of tumour growth-promoting genes, respectively.11–14 For example, unsaturated fatty acids inhibit the expression of phosphatase and tensin homologue (PTEN) in hepatocytes via activation of an NF-κB/mammalian target of rapamycin (mTOR) complex.15 16 PTEN is a regulator of phosphoinositide 3-kinase (PI3K) signalling and an important tumour suppressor mutated/deleted in HCC, and consistently, HCC growth and progression were significantly increased in a xenograft model of mice fed with an oleic acid-enriched diet, even without weight gain.15

Similarly, Park et al recently showed that feeding a high-fat diet (HFD) promoted HCC formation in a diethylnitrosamine-induced HCC model.17 Interestingly, JNK and mTOR activity were increased in HFD-fed mice, and the activity of pro-oncogenic pathways as extracellular signal-regulated kinase (ERK) and signal transducer and activator of transcription 3 (STAT3) was upregulated.17 STAT3 is a major transcriptional target of interleukin 6 (IL-6) and its activation is a typical feature in human HCCs18; mice deficient in IL-6 were protected from the tumour promoting effect of HFD in the diethylnitrosamine-induced HCC model. Moreover, mice with deficient tumour necrosis factor (TNF) were also protected from obesity-enhanced hepatocarcinogenesis. TNF is a strong inducer of pro-oncogenic pathways, including NF-κB, and notably, obesity induced TNF and IL-6 levels in both non-tumorous liver and HCC.19

One further example of signalling pathways/molecules similarly regulated in NAFLD and HCC are sterol regulatory element binding proteins (SREBPs), master regulators of hepatic lipogenesis, which are activated early in response to hepatic steatosis. In HCC, lipogenesis is markedly induced via SREBP1 and correlates with a poor prognosis.20 Interestingly, epitopes of hepatitis B and C virus (HBV, HCV) may also elicit hepatic steatosis via induction of SREBP1,21 22 and patients chronically infected with HCV and fatty liver carry a higher risk of HCC than those with either condition alone.23 These data suggest that hepatic lipid accumulation ‘sensitises’ (metabolic) pathways that predispose to malignant transformation already at early stages of NAFLD and eventually, in chronic liver disease itself. A tumour-promoting environment is further augmented by mitochondrial dysfunction and increased free fatty oxidation in peroxisomes and microsomes, which lead to an excess of free reactive oxygen species and lipid peroxides which can form adducts with DNA, potentially driving normal cells into malignancy.

Another important mechanism by which oxidative stress may induce hepatic malignancy is dysfunction of the endoplasmic reticulum (ER). Accumulating evidence suggests that fatty acids can induce ER stress and ER stress-associated apoptosis in various metabolic cells by triggering calcium signals and free radicals.24 These data suggest a crucial role for the ER in the pathogenesis of NAFLD, where high intracellular lipid stores charge the ER with a consecutive increase of oxidative stress, inflammation and activation of oxidative pressure-sensitive intracellular survival pathways, including NF-κB and JNK.

In addition, hepatic iron overload frequently seen in patients with NAFLD contributes to the generation of oxidative stress,25 and is known to increase the risk of HCC in NASH-induced cirrhosis.26 Together, these studies suggest that fatty liver must be viewed as a risk factor for HCC evolution independent of other pathophysiological components of the MS.

Moreover, hepatic lipid accumulation causes IR both locally in the liver and systemically through NF-κB activation.27 Here, hepatic steatosis is not only one feature of the MS but actively enhances IR as its underlying pathophysiological mechanism. IR leads to hyperinsulinaemia (or vice versa), which activates PI3K/Akt, implicated in hepatocarcinogenesis. Thus, it seems fair to speculate that cellular survival advantages develop in the liver along with IR and thereby contribute to NAFLD-associated HCC.

Besides the liver, other organ dysfunctions related to the MS may indirectly affect HCC progression. Thus, obesity leads also to significant changes of the expression and secretion of adipose tissue-derived adipokines such as leptin. Leptin is significantly increased in patients with NAFLD, and promotes angiogenesis and the progression from NASH to HCC in mice.28 Conversely, adiponectin is decreased in patients with NAFLD, hypoadiponectinaemia favours tumourigenesis in a murine NASH model29 and adiponectin inhibits HCC growth and metastasis by suppression of tumour angiogenesis.30 Together these findings underline the complexity of the interaction of hepatic and extrahepatic pathophysiology, jointly promoting HCC in NAFLD or MS, respectively. In addition, systemic alterations present in the MS synergistically drive progression of NAFLD from steatosis to NASH and eventually, cirrhosis, which provides the soil for malignant transformation and cancer progression. Figure 1 depicts the mechanisms linking MS and NAFLD with HCC.

Figure 1

Molecular mechanisms linking non-alcoholic fatty liver disease (NAFLD) with the development of hepatocellular carcinoma (HCC). Obesity and the metabolic syndrome reached a pandemic dimension, and most of these individuals have fatty liver. The evolution of NAFLD requires the concomitance of (relative) energy oversupply and acquired or inherited insulin resistance. Elevated levels of proinflammatory cytokines as tumour necrosis factor (TNF) and interleukin 6 (IL-6) or the adipokine leptin, and decreased anti-inflammatory signals (adiponectin) activate molecular pathways promoting inflammation and cell survival. The same pathways are also known as oncogenic signals. Concomitantly, apoptosis and the tumour suppressor phosphatase and tensin homologue (PTEN) are suppressed. Furthermore, excess fatty acid supply and steatosis elicit increased fatty acid oxidation with consecutively enhanced reactive oxidative stress (ROS) from microsomal enzymes and endoplasmic reticulum (ER) dysfunction. ER-stress and ROS further promote proinflammatory and pro-oncogenic signals. Moreover, they provoke malignant transformation also through generation of reactive aldehydes. Importantly, these cascades of events may take place in the absence of cirrhosis. On the other hand, pathophysiological alterations promote also hepatic inflammation and fibrosis and also, indirectly, tumourigenesis in NAFLD.

Clinical and socioeconomic implications of the NAFLD–HCC connection

While the absolute risk for HCC in NAFLD may be relatively low, the rapid increase of the MS and its hepatic manifestation would suggest a modest increase in risk in a large total number of HCC cases. Bearing this in mind it is alarming that the incidence of HCC is decreasing in high-prevalence areas of the world, whereas its incidence in low-prevalence regions such as Europe and the United States has nearly doubled.31 While the former is possibly the result of large-scale vaccination against HBV infection and decreased exposure to dietary aflatoxins, the latter may be, at least in part, attributable to the rising incidence of NAFLD/NASH. Still, the dramatic increase of obesity is relatively recent, and it takes several decades before NASH develops into cirrhosis. Under these circumstances, the rising incidence of severe obesity in children is particularly worrying and concerns seem justified considering the relatively poor outcome of HCC in NAFLD compared with other liver diseases.

Parts of the problem are related to coexisting vascular diseases, but also to unfavourable tumour characteristics owing to delayed diagnosis in patients not commonly included in surveillance programmes. Thus, case series of patients with HCC with NAFLD as the only risk factor showed that half of the HCCs were detected at the time of first referral, and in several of the remaining cases the time lag between the diagnosis of liver disease and HCC detection was only 6 months, suggesting a serious lack of previous HCC screening measures.4 Consequently, patients with HCC in cryptogenic cirrhosis more often had larger and multifocal tumours, and were less frequently amenable to surgical resection or local ablation.32 33 In addition, an increased risk of HCC recurrence after local ablation or surgical resection was noted in patients with visceral obesity and diabetes, respectively.34 35

Prospective evidence demonstrating the evolution of HCC in NAFLD derives from two larger studies comparing the incidence of HCC in patients with NAFLD versus those with HCV-related cirrhosis. In the study of Hui et al, none of the 23 patients with NAFLD developed HCC, while eight of 46 patients with HCV did.36 However, the overall survival and complication rate were similar in both groups. The other study compared 152 patients with NAFLD with 150 matched subjects with HCV-associated cirrhosis and noted a similarly poor outcome in end-stage cirrhosis.37 These studies suggest that HCC is less common among patients with NAFLD than among those with HCV cirrhosis, but the prognosis is equally grave as liver disease decompensates. Although studies on HCC evolution in patients with NAFLD are difficult to perform owing to its long, asymptomatic and predominantly benign course requiring large numbers of subjects and long-term surveillance, this type of epidemiological data is urgently needed to gain an estimate of the problem's true dimensions.

Obviously, improved surveillance is warranted, but a lack of predictors indicating an increased risk of developing HCC in NAFLD poses difficulties in devising evidence-based screening strategies: should we screen all patients with histologically confirmed NASH regardless of the stage of fibrosis? Or even all patients with NAFLD as assessed by ultrasound? Although ultrasound has an acceptable sensitivity, detecting hepatic steatosis at a reliable threshold of approximately 30% fat,38 only individuals with significant steatosis would be identified. Moreover, ultrasound fails to separate patients with benign steatosis from those with significant necroinflammation, who are the ones at high risk for progression of NAFLD to complicated NASH according to a recent meta-analysis.39 Therefore, liver biopsy remains the diagnostic cornerstone for identifying patients with progressive NASH, and could potentially become, by means of yet to be defined histological diagnostics, the crucial examination to distinguish between those who may develop HCC and those who probably will not. Potentially, the diagnostic power of histology may by enhanced by specific analysis of certain molecular markers.40

Furthermore, great expectations are linked with genome-wide association studies (GWAS) that allow for high through-put analysis of genetic variants to detect those that may predispose for a certain disease. Interestingly, a recent GWAS found a significant association of a variant in the gene coding for PNPLA3 (adiponutrin) with NAFLD41; however, no data exist showing whether this polymorphism also predisposes for carriers that develop HCC. Clearly, medical capacities and financial resources would be overstretched considering the close association of NAFLD with obesity and diabetes, and the high prevalence of these conditions, if no additional markers can be defined that allow for focused screening of those at risk for HCC.

With this dilemma, HCC prevention becomes an even more important concern than with other chronic liver diseases associated with HCC where surveillance and causative treatments are established. However, until now no studies have shown that any of the interventions–including drug treatments–specific for NAFLD can prevent HCC. Experimental studies indicate that statins have a potential cancer prevention effect, and a recent study by El-Serag et al showed that statin use is associated with a significant reduction in the risk of HCC among patients with diabetes.42 However, the use of statins is influenced by the socioeconomic status, which also affects cancer risk factors such as poor nutrition or physical activity.43 Randomised, prospective trials are required to confirm that hepatocarcinogenesis is directly affected by statins independent of confounding effects.

Besides pharmacological interventions for treating IR, the most obvious and straightforward strategies appear to be those leading to weight loss. Here, it appears intriguing that dietary guidelines may have to include recommendations about consumption of specific types of fat, since several studies showed different effects of saturated and unsaturated fats on cancer (related pathways).11 14 Considering the obvious reluctance of many obese people to reduce their calorie intake, changes of type of nutrient rather than reduction of the amount of nutrients might be more suitable for some. Thus, the most recent studies, which indicate a protective role of coffee for both NAFLD and HCC, are of high interest.44 45 However, prospective nutritional studies must control for confounders before dietary recommendations can be given.

Finally, the role of alcohol drinking in NAFLD is important, since there are few patients with NAFLD who entirely abstain from drinking while clearly more common is the ‘obese drinker’. In general, NAFLD/MS and additional liver insults exhibit confounding effects on hepatic steatosis, inflammation, fibrosis as well as on carcinogenesis. Thus, NAFLD lesions sufficient to fulfil the criteria for NASH can be found in patients with chronic alcohol abuse or chronic HCV infection, thereby accelerating their natural progression to severe liver damage. Accordingly, obesity is an independent risk factor for the progression of both alcoholic and viral liver disease, and diabetes is a synergistic risk factor for virus-mediated and alcohol-induced HCC.46 Therefore, regardless of its ‘own’ impact on hepatocarcinogenesis, fatty liver may be the last straw that breaks the camel's back in patients with other chronic liver disease. Consequently, dietary recommendations or tight diabetic control for tumour prevention should not be restricted to patients with ‘pure’ NAFLD but should be expanded to patients with other liver disease and accompanying MS.

Perspective and conclusions

There is clear epidemiological evidence and substantial biological plausibility that NAFLD is a risk factor for HCC. It promotes both HCC development and progression, and in addition to hepatic steatosis, further pathophysiological features of the MS seem to have an important role. The prevalence of NAFLD is increasing owing to a lifestyle favouring the development of the MS, and thus, even if the absolute risk for HCC in NAFLD may be relatively low, it can be expected that HCC with the background of NAFLD will contribute significantly to the disease burden derived from HCC in the future. Currently, there are no established risk factors or biomarkers and no intervention strategies that guide the management of HCC in NAFLD. Apart from unresolved problems in the pathophysiology of NAFLD and the molecular carcinogenesis of HCC in this context, more epidemiological information is needed to assess the true incidence of HCC during the natural course NAFLD and to identify patients at risk. Subsequently, this information must be translated into implementing preventive measures such as HCC screening and interventions that effectively stop HCC development.


View Abstract


  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.