The healthcare ‘maths’ of liver-disease diagnosis and treatment simply doesn’t add up, not only in terms of the large and increasing cost to healthcare systems globally, but also – and far more importantly – from the perspective of poor individual health outcomes.
And the main factor in the equation is the failure of the current diagnostic paradigm to detect disease before it’s too late.
Chronic liver disease is one of the leading causes of human mortality globally, resulting in almost two million deaths per annum, due largely to lifestyle factors in developed economies (metabolic disease, alcohol consumption), and from viral hepatitis in emerging markets (World Health Organization, Global Burden of Disease 2010). Recent estimates suggest that as many as 50% of the adult population in some Western countries have undiagnosed non-alcoholic fatty liver disease (NAFLD), which is the precursor to more serious chronic liver disease (see Figure). Yet this cluster of diseases remains significantly under-addressed in global healthcare responses, resulting in one of the largest sources of preventable death in the world.
The most significant hurdle is that the ‘gold-standard’ for diagnosis of liver-related disorders is a surgical procedure – liver biopsy and histopathology. While this enables clinicians to identify other liver injuries – such as inflammation, steatosis or necrosis – it is not only invasive, painful and expensive, but it is often inaccurate due to sampling errors (e.g. clinician taking tissue from a non-diseased section of liver) and there is significant variability in interpretation of pathology results. Widespread use of an invasive surgical procedure to screen for disease is, obviously, infeasible – and this is particularly problematic as the early stages of liver disease are often asymptomatic, and disease diagnosed at a late stage is much harder to treat successfully.
A number of non-invasive tests for liver disease are available, but none has yet to supplant biopsy for confirmatory diagnosis. So-called ‘liver-function’ tests based on blood biomarkers such as liver transaminases, albumin and bilirubin are widely used, but show generally poor performance: a recent community-based, prospective study of patients with abnormal liver-function results established diagnosis in fewer than 5% of cases (Lilford et al., 2013).
In general, biomarkers that perform well as diagnostic tests are those that closely reflect underlying pathophysiology and can reveal disease staging and progression so that differential clinical decisions can be made. In this regard, the key factors in diagnosis of liver disease are determining the degree and rate of progression of liver fibrosis, although markers of inflammation and necrosis do provide distinct information (see Figure). Novel imaging methods such as transient elastography are useful for detecting fibrosis, but do not provide information on the underlying disease processes. A significant focus of discussion at the International Liver Convention in London this past April was around ways in which such imaging methods could be combined with new (more relevant) blood-based biomarker tests/panels to provide more sensitive diagnosis.
However, despite many of these newer tests providing improved information, they remain reliable only for late-stage diagnosis, and determination of progression is simplistic and can result in errors of diagnosis (both false-positives and negatives). This is because, currently, tests compare biomarker levels for different stages of liver fibrosis using pre-determined cut-offs based on population averages to separate the stages. However, some people may show different biomarker levels simply because of natural healthy variation and might therefore be diagnosed incorrectly. Further, repeat testing over time and ‘recategorisation’ on the basis of changes in biomarker levels is ad hoc and remains prone to miscategorisation for the precisely same reason – the cut-offs are not personalised.
And this is where the mathematics of diagnosis comes in to its own: measuring biomarkers in individuals at regular intervals when healthy (serial screening) enables personalisation of a diagnostic by spotting changes in a marker relative to their own baseline, rather than in comparison to other people’s levels. Quantitative algorithms can be used to identify even small biomarker changes, enabling a far more powerful method of diagnosis (increased sensitivity/specificity), meaning that relevant changes can be picked up far earlier in disease progression, and enabling intervention and improved healthcare outcomes.
October is Liver Awareness Month.
Dr Simon Goldman and
Dr Wendy Alderton
Abcodia Limited