The hepatitis C virus (HCV) was discovered in the late 1980s, and a decade later, drugs became available which could eradicate the virus in almost half of infected patients. Quantitation of the hepatitis B virus (HBV) through sensitive commercial assays became possible only a few years ago; today, a wide range of powerful antiviral drugs is available. In sharp contrast, nonalcoholic steatohepatitis (NASH), which was first described a quarter of a century ago,1 became a subject of intense scrutiny and publication in the following decades2, 3 and although its disease spectrum keeps expanding,2, 4, 5 the overall perception in the majority of cases is still that of a benign disease. The prognosis of nonalcoholic fatty liver disease (NAFLD) has recently been rated from remarkably benign,6 unimpressive,7 very unimpressive8 to the use of the “I” word (incidentaloma).9 Of particular concern is that most patients at risk for NAFLD are not primarily referred to hepatologists but to specialists in the endocrine or nutrition fields who seem to endorse an optimistic view of hepatic steatosis and its consequences. Intriguingly (and sadly enough), no single drug development program in NASH is currently underway, while drugs designed for weight control or diabetes treatment are, by and large, available and in fierce competition among pharmaceutical companies.

When contemplating this paradox, one might first consider that, even in liver diseases already established as leading causes of liver failure (such as chronic hepatitis B or C, alcoholic liver disease or hemochromatosis), the assessment of prognosis has all been controversial.10, 11 Different views and seemingly contradictory observations stem from the fact that most liver diseases have considerable inter-individual variability in presentation and natural course. HCV infection is a good example: depending on the age of infection, sex, mode of transmission and comorbid associations,12 the disease could take more than a lifetime before evolving to cirrhosis13 or it could undergo accelerated fibrogenesis14 with a heavy impact on short-term liver-related mortality.15 HBV infection, although a leading cause of mortality and liver cancer in certain parts of the world, has been shown under certain circumstances to evolve as a benign disease with no mortality or disease decompensation in large cohorts of patients followed for more than 30 years.16 Overall, 20% or less of patients with an alcohol intake considerably above the one associated with risk of liver damage will ever go on to develop liver cirrhosis. To add to this short list, it becomes increasingly unclear whether in genetic hemochromatosis, iron overload alone could even result in liver cirrhosis in the absence of other cofactors such as alcohol, obesity, or HCV infection.

In NAFLD, this potential for individual heterogeneity is also present, only exacerbated to a higher degree. Fat accumulation in the liver seems to be a very early event in the course of insulin resistance. As a result, different profiles of patients with this disease are encountered in clinical practice. These range from lean individuals with minimal excess truncal fat to those with morbid obesity, or from patients with normal glycemic control to those with full blown, type 2 diabetes with metabolic complications. Unfortunately, this also makes it almost impossible to perform adequate clinical and prognostic studies over the entire spectrum of liver injury, for several important reasons which are specific to NAFLD. First, most patients with NAFLD, and probably almost all of those with normal aminotransferases are not referred to hepatologists (the opposite being true for patients diagnosed with viral hepatitis infection) due to the perceived benignity of steatosis and its silent presentation. Second, there is a regrettable lack of a simple and specific diagnostic marker for this disease, and therefore screening of large populations through the current gold standard, liver biopsy, is impossible, let alone that non-hepatologists are not at ease with the procedure. Hence, observations from different populations will always suffer from ascertainment bias and underreporting silent NAFLD on a population-based level is inevitable and is expected to keep feeding controversy. A study with perfect methodology addressing the prognostic impact of NAFLD in the general population would certainly be very welcome, but realistically would be too complex to perform in the foreseeable future. In the meantime, we should not overlook the accumulated knowledge which points towards a more severe disease than was originally thought.

What is the current knowledge of the clinical outcome of NAFLD? Bland steatosis or steatosis with minimal inflammation insufficient to qualify as NASH is most likely a nonprogressive disease when occurring alone.17, 18 It is uncertain whether the marginal proportion of cirrhosis occasionally reported19, 20 reflects a real, albeit rare, propensity to fibrogenesis or simply missed lesions of NASH at the initial liver biopsy.21 When steatosis occurs in association with another liver disease, however, the situation might be different as this seems to favor progression of fibrosis22; in this case, the distinction between bland steatosis and steatohepatitis is even harder to make with confidence. A wealth of experimental data demonstrate a particular vulnerability of fatty liver to numerous superimposed noxious agents.23, 24 Caution is in order before concluding that bland steatosis is benign.

Steatohepatitis on the other hand, can result in advanced liver fibrosis and cirrhosis25, 26 in a significant proportion of cases. From the hepatologist's perspective, NASH is an important cause of advanced liver fibrosis in clinical practice.27 In a recent multicentric survey of 272 French patients with chronic unexplained hypertransaminasemia,28 90% of cases of cirrhosis were found in patients with steatohepatitis. In fact, the importance of NASH as a cause of cirrhosis is underestimated, as the diagnosis of NASH at this stage relies more on exposure to cardiometabolic risk factors (overweight, diabetes)29 than on histological features. Findings from the large UNOS database show a step-wise increase between the proportion of cryptogenic cirrhosis and increasing BMI, with cryptogenic cirrhosis being more frequent in diabetics than in non-diabetics.30

The hepatologist's perspective on the importance of NASH can be challenged due to considerable selection bias. However, an important body of evidence from large unselected cohort studies with long follow-up remind us that obesity and diabetes, two strong prosteatogenic conditions, can induce end-stage liver disease and its complications. After a mean follow-up of 13 years, obese, nondrinking individuals from the general population had a 4-fold increased risk of dying from cirrhosis or of developing its complications when compared to lean patients.31 Diabetic patients have at least a 2.5-fold higher risk of dying from cirrhosis than the general population32 (again, a conservative figure considering that some complications of cirrhosis such as infections can easily be classified as unrelated). Long-term follow-up data from a case-control study of diabetic males with a huge sample size have shown a time-dependent increase in the risk of developing NAFLD compared to non-diabetics: a 36% increase in those followed less than 5 years and a 2-fold increase in those followed more than 10 years.33 Granted, most of these studies did not exclude other liver diseases, such as viral hepatitis, with state of the art molecular diagnostic methods. But it is highly improbable that the bulk of liver disease in these patients is explained by causes other than NASH, given the strong epidemiological association between NAFLD and insulin resistance. Moreover, numerous studies have now shown that in countries with both low34–38 and high39, 40 prevalence of viral hepatitis, diabetes increases by 2- to 5-fold the risk of another complication of cirrhosis, namely hepatocellular carcinoma. Indeed, there is a consistent, dose-response correlation between fasting and 2h serum insulin with the risk of death from primary liver cancer in nondiabetic French men41 as well as a stepwise increase between fasting glucose levels and the incidence and risk of death by liver cancer in Korean men.40

A frequent liver disease that results in cirrhosis and liver cancer should be sufficient as a cause of great concern. However, arguably, only the demonstration of increased overall and liver-related death could remove any reasonable doubt. Data on this aspect are slowly emerging in the field of NAFLD. NASH-induced cirrhosis reduces survival and predisposes to decompensation of cirrhosis just as any other etiology.42 An important distinction, however, is that decompensation occurs slowly in Child A cirrhosis, while in Child B or C, once complications have occurred, the prognosis is particularly grim.42 This has been recently confirmed by Sanyal et al., who further demonstrated that causes of death are liver-related and predictors of survival are basically the same in NASH-induced cirrhosis as in any other cirrhosis (MELD score and renal function).43 In the end, whether the outcome of NASH-induced cirrhosis is better or worse than that of HCV-induced cirrhosis42–44 does not really matter and will be confounded anyway by lead-time bias due to current limitations in the diagnosis of NASH, as outlined above. Rather, we should be concerned about how to diagnose NASH-induced advanced fibrosis in high-risk patients and prevent its progression to cirrhosis, as once decompensation occurs, little can be done to avoid a fatal outcome in these patients who, for the most part, are weaker candidates for liver transplantation.42, 43 A more systematic attempt at reducing selection bias evaluated the impact of NAFLD-related mortality at a population-based level.18 The results of this important study by Adams et al. are, again, an eye-opener: patients with NAFLD are reported to have an excess mortality risk of 34% over the general age and sex-matched population after a mean follow-up of only 7.6 years which, coherently, increased to 55% with a longer follow-up (10 years). True, this excess mortality is not entirely attributable to liver failure. But cirrhosis was the third leading cause of death in the NAFLD cohort versus the 13th in the control population (which confirms previous findings19) with a median survival shorter than 7 years.18

These data have nonetheless fallen short from settling any controversy and fuel further debate. It has been argued that since mortality occurs with cirrhosis, excluding these patients from the survival analysis will show that NAFLD is in fact much more benign than previously thought.7 However, all chronic liver diseases share one common feature, which is that mortality occurs at the cirrhotic stage and not before. Excluding such patients from the survival analysis will make chronic HBV and HCV hepatitis look just as “benign” as NAFLD. And there is no reason to exclude such patients when assessing the severity of a disease that can induce cirrhosis: Adams et al. noted that cirrhosis occurred during follow-up in 13 of their 21 cases.18 Failure to demonstrate impaired survival in noncirrhotic NAFLD does not mean NAFLD is benign. It simply means that follow-up was too short to capture the development of cirrhosis and the occurrence of its complications; this comes as no surprise as this typically takes 20 to 30 years in chronic hepatitis C.13

In line with these arguments, an important addition to the field is made by Ekstedt et al. in this issue of HEPATOLOGY.45 The authors report on the long term clinical and histological follow-up of a consecutive cohort of NAFLD patients initially referred for investigation of abnormal liver function tests. Information on clinical status was available in all 129 patients and repeat liver biopsy in 68, after a mean follow-up of 13.7 years. Importantly, the authors were able to identify the pathological form of NAFLD, NASH or simple steatosis, at inclusion and hence study the natural course based on this distinction. Patients with NASH had significantly reduced survival compared to the age and sex-matched general population and a significantly higher risk of liver-related (2.8% vs. 0.2%, respectively) and cardiovascular-related death (15.5% vs. 7.5%, respectively). End-stage liver disease (ESLD) occurred in 10% during follow-up, including three cases of hepatocellular carcinoma. Remarkably, and possibly because of the longer follow-up, the authors were able to show that even patients without cirrhosis developed ESLD (18%, 6/34 F2 and F3 patients) during follow-up, thus refuting the claim that noncirrhotic NASH is “benign”. In contrast, patients with steatosis but no NASH had similar survival and similar causes of death as the general population and displayed no ESLD. These findings confirm and extend previous results20, 46 as well as acknowledge the considerable potential for heterogeneity of the clinical course in NAFLD and the need to incorporate the distinction between its different pathological forms in future studies. Another strength of the study by Ekstedt et al.45 is the long follow-up period, which extended well beyond the first decade after diagnosis. This allowed the authors to document a significant proportion of progression of fibrosis on repeat liver biopsy: 15 out of the 66 (23%) patients with F0 to F2 fibrosis on initial biopsy progressed by 2 or more stages and 6 developed cirrhosis during follow-up.

Last but not least, the study by Ekstedt et al., together with previously published data43 point to an intriguing finding that might be critical for future research: the possibility of increased cardiovascular mortality in patients with NASH. Several studies have shown increased risk scores for cardiovascular events (such as the Framingham score) in patients with NASH,47 even after adjustment for BMI, age and other known risk factors.48, 49 Others have demonstrated a higher prevalence of carotid plaques,50, 51 endothelial dysfunction47 or higher carotid media thickness in patients with NAFLD.51, 52 A great challenge will be to specifically evaluate the additional risk conferred by NASH, independent of the underlying metabolic abnormalities.

Areas of uncertainty regarding the impact of NAFLD on a population-based level exist, and legitimate questions are still unanswered. However, it has become clear that for patients we see in liver clinics, NASH is a disease with serious fibrogenic potential which can result in liver-related morbidity and mortality. As such, it should be regarded as a major unmet medical need. Strategies for screening patients with cardiometabolic risk factors for liver injury should be implemented and large therapeutic trials should no longer be delayed. As far as NASH goes the “I” word should neither be “incidentaloma” nor “inaction”.


  1. Top of page
  2. References