Advertisement

Predicting the future burden of NAFLD and NASH

  • Suzanne E. Mahady
    Affiliations
    School of Public Health & Preventive Medicine, Monash University, Clayton, Australia

    Department of Gastroenterology, Royal Melbourne Hospital, Parkville, Victoria, Australia
    Search for articles by this author
  • Jacob George
    Correspondence
    Corresponding author. Address: Department of Medicine, Westmead Hospital, Westmead, NSW 2145, Australia. Tel.: +61 2 88907705; fax: +61 2 88907582.
    Affiliations
    Storr Liver Centre, The Westmead Institute for Medical Research, Westmead Hospital and University of Sydney, NSW, Australia
    Search for articles by this author
      While the prevalence of non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH) continues to rise, with both conditions increasingly recognised in primary care and specialist practice, quantifying future disease burden has always been challenging. This is largely due to the diagnostic tools currently available for NAFLD diagnosis. Ultrasound is poorly sensitive for milder forms of hepatic steatosis, while liver enzymes are variably elevated in NAFLD and may be normal in 50–80% of cases. Even the ‘gold’ standard of liver histology has a substantial measurement error.
      • Ratziu V.
      • Charlotte F.
      • Heurtier A.
      • Gombert S.
      • Giral P.
      • Bruckert E.
      • et al.
      Sampling variability of liver biopsy in non-alcoholic fatty liver disease.
      Compounding these issues is the lack of a standardised approach in NAFLD prevalence studies globally. Studies vary in the populations sampled (primary vs. tertiary care), the diagnostic tool used (ultrasound or enzymes), or have only considered one point in time. Accurate prevalence data is essential to inform modelling studies of NAFLD disease burden, which in turn is needed to support healthcare policy and planning.
      In this issue of the Journal, Estes and colleagues have sidestepped this problem with an innovative approach. Using population based, epidemiological data on obesity and type 2 diabetes from national health surveys, they have applied the expected prevalence of NAFLD and NASH in these conditions to provide estimates of prevalence at a population level. These data, along with figures obtained from a systematic review of prevalence studies, have been used in a Markov model to approximate the population-level prevalence of NAFLD and to improve precision compared to inclusion of data from NAFLD prevalence studies alone. As a diagnostic test for obesity, body mass index is reasonably accurate when compared against a reference standard of body fat percentage-defined obesity
      • Romero-Corral A.
      • Somers V.K.
      • Sierra-Johnson J.
      • Thomas R.J.
      • Collazo-Clavell M.L.
      • Korinek J.
      • et al.
      Accuracy of body mass index to diagnose obesity in the US adult population.
      and is strongly associated with NAFLD, although the association between NAFLD and truncal obesity is more robust and would have improved precision further.
      • van der Poorten D.
      • Milner K.L.
      • Hui J.
      • Hodge A.
      • Trenell M.I.
      • Kench J.G.
      • et al.
      Visceral fat: a key mediator of steatohepatitis in metabolic liver disease.
      Nevertheless, this is a creative approach to approximating NAFLD prevalence, and ongoing innovation in the absence of high quality studies remains important.
      The current modelling study is also more ambitious in scope than previous efforts. NAFLD prevalence and incidence was modelled from 2016 to 2030, in eight countries accounting for a quarter of the world’s population. This includes the United States, United Kingdom, Germany, Italy, France, Spain, China and Japan. To estimate future cases of NAFLD and NASH, the authors assigned estimated baseline fibrosis stage to the modelled population prevalence, adjusted for age and sex, and modelled fibrosis progression to 2030 using estimates from histological studies.
      • Singh S.
      • Allen A.M.
      • Wang Z.
      • Prokop L.J.
      • Murad M.H.
      • Loomba R.
      Fibrosis progression in non-alcoholic fatty liver vs non-alcoholic steatohepatitis: a systematic review and meta-analysis of paired biopsy studies.
      For mortality estimates, an additional increment in cardiovascular mortality of 1.15 was applied, in acknowledgment of data suggesting an excess risk of cardiovascular mortality in people with NAFLD,
      • Targher G.
      • Byrne C.D.
      • Lonardo A.
      • Zoppini G.
      • Barbui C.
      Nonalcoholic fatty liver disease and risk of incident cardiovascular disease: a meta-analysis of observational studies.
      although other evidence shows no increase.
      • Stepanova M.
      • Younossi Z.
      Independent association between non-alcoholic fatty liver disease and cardiovascular disease in the general population.
      The salient finding of the model is a significant rise in cases of NASH and advanced fibrosis (more so than NAFLD alone) by 2030, mainly because of the effect of ageing within the cohorts. The largest population, China, was predicted to have the greatest overall number of people with NAFLD (estimated at 314 million by 2030), but the U.S. had the fastest rising incidence of advanced disease. The greatest incidence increase in advanced liver fibrosis was expected to occur in countries with older populations like France and Germany, with a modest number of liver-related deaths by 2030, estimated at 4,530 in Spain and 7,030 in France. However, prediction of liver-related mortality in older people is fraught with ambiguity, as older people have competing risks of mortality from cardiovascular disease and cancer as the cohort ages, and there is insufficient data on liver-related outcomes in elderly populations with NAFLD to adequately inform projections.
      All models are inherently uncertain. Assumptions need to be made where data is absent, expert opinion alone informs important parameters, and error in input values may substantially impact the results when exponentially modelled. Estes and colleagues have built a credible model with the best available data and reasonable assumptions, but there are a number of important areas where differences in expert opinion would likely change the current predictions. For example, while obesity is viewed by some as an unremitting condition, some US based population studies suggest a levelling off of global obesity prevalence.
      • Flegal K.M.
      • Kruszon-Moran D.
      • Carroll M.D.
      • Fryar C.D.
      • Ogden C.L.
      Trends in obesity among adults in the United States, 2005–2014.
      Global drivers of obesity may change (for example, the introduction of a sugar tax), while more effective interventions to reduce obesity may be developed. Fibrosis progression rates remain largely undefined, and more accurate data in this area would significantly improve precision in future modelling studies. It is unlikely that fibrosis progression rates are the same in different age groups and ethnic populations, and as for hepatitis C, these are likely to vary with time. To this end, the development of validated non-invasive tests or algorithms that reflect liver fibrosis stage (either compared to reference standards of liver biopsy or magnetic resonance elastography) and can be used for population-level screening, is an important but unmet clinical need. Additionally, lean NAFLD was not included in the model but is an important consideration in Asian populations.
      Uncertainty in a model’s results should be explored with sensitivity analyses, where a plausible range of values for key parameters are tested to determine if the results change to a clinically meaningful extent. In this model, NAFLD prevalence and the standardised mortality ratio account for most of the uncertainty. It would have been ideal if sensitivity analyses and their results in these areas in particular were included and discussed in the main manuscript, rather than supplementary data. This gives both readers and researchers a clearer understanding of the robustness of the model and its estimates, and may inform directions for data collection in future models. In addition, point estimates for cases are best expressed using confidence intervals, as the latter is more likely to contain the accurate figure. Despite these limitations, the current model is a comprehensive study, and it is desirable that this should be viewed as a 'living document', amenable to inclusion of emerging data, so that the immense intellectual effort may be capitalised upon.
      Finally, modelling studies are perhaps most valuable in their ability to highlight important evidence gaps for research prioritisation. In NAFLD, there remains a paucity of data in important areas, mandating an international, concerted effort to improve our understanding. Trials and observational studies with clinical end points, rather than the more easily measured surrogate end points which are often misleading, are essential.
      • Fleming T.R.
      • DeMets D.L.
      Surrogate end points in clinical trials: are we being misled?.
      This is particularly true when we have little data on how well surrogates like fibrosis stage truly correlate with disease outcomes. Fibrosis progression rates are also subject to measurement error and are highly variable, particularly in NAFLD which requires a systems approach to understand both pathogenesis and progression. Researchers planning NAFLD prevalence studies should use a range of measures and technologies to increase accuracy, and ideally have an in-built research biobank that can be interrogated when new diagnostic or prognostic tools emerge. These are the building blocks needed to inform a proper burden of disease study using DALYs (disability adjusted life years), that will provide a more comparable analysis of the true burden of NAFLD and will be a powerful tool to inform health policy.

      Financial support

      The authors received no financial support to produce this manuscript.

      Conflict of interest

      The authors declare no conflicts of interest that pertain to this work. Please refer to the accompanying ICMJE disclosure forms for further details.

      Supplementary data

      References

        • Ratziu V.
        • Charlotte F.
        • Heurtier A.
        • Gombert S.
        • Giral P.
        • Bruckert E.
        • et al.
        Sampling variability of liver biopsy in non-alcoholic fatty liver disease.
        Gastroenterology. 2005; 128: 1898
        • Romero-Corral A.
        • Somers V.K.
        • Sierra-Johnson J.
        • Thomas R.J.
        • Collazo-Clavell M.L.
        • Korinek J.
        • et al.
        Accuracy of body mass index to diagnose obesity in the US adult population.
        Int J Obesity. 2008; 32: 959
        • van der Poorten D.
        • Milner K.L.
        • Hui J.
        • Hodge A.
        • Trenell M.I.
        • Kench J.G.
        • et al.
        Visceral fat: a key mediator of steatohepatitis in metabolic liver disease.
        Hepatology. 2008; 48: 449
        • Singh S.
        • Allen A.M.
        • Wang Z.
        • Prokop L.J.
        • Murad M.H.
        • Loomba R.
        Fibrosis progression in non-alcoholic fatty liver vs non-alcoholic steatohepatitis: a systematic review and meta-analysis of paired biopsy studies.
        Clin Gastroenterol Hepatol. 2015; 13: 643
        • Targher G.
        • Byrne C.D.
        • Lonardo A.
        • Zoppini G.
        • Barbui C.
        Nonalcoholic fatty liver disease and risk of incident cardiovascular disease: a meta-analysis of observational studies.
        J Hepatol. 2016; 65: 589
        • Stepanova M.
        • Younossi Z.
        Independent association between non-alcoholic fatty liver disease and cardiovascular disease in the general population.
        Clin Gastroenterol Hepatol. 2012; 10: 646
        • Flegal K.M.
        • Kruszon-Moran D.
        • Carroll M.D.
        • Fryar C.D.
        • Ogden C.L.
        Trends in obesity among adults in the United States, 2005–2014.
        JAMA. 2016; 315: 2284
        • Fleming T.R.
        • DeMets D.L.
        Surrogate end points in clinical trials: are we being misled?.
        Ann Intern Med. 1996; 125: 605

      Linked Article