Background

Non-alcoholic fatty liver disease (NAFLD) is defined as fatty infiltration of the liver in the absence of alcohol misuse or other causes of liver damage and is probably the most common liver disorder in all ages in the developed world today. NAFLD includes a wide spectrum of liver damage, ranging from simple steatosis to advanced fibrosis and cirrhosis (this advanced form referred to as non-alcoholic steatohepetitis (NASH)). The primary cause of NAFLD is obesity and with the global epidemic of obesity the prevalence of NAFLD and its complications are increasing. NAFLD has recently been described as "a disease of our generation", as clinicians have noted a large increase in liver pathology in all ages driven by the obesity epidemic, in addition to increasing rates of alcohol consumption.

The whole spectrum of NAFLD has been described in children and adolescents. Presentation with, and progression to, cirrhosis in adolescence has also been described. Estimates of the prevalence of NAFLD, based on unexplained elevated levels of alanine aminotransferase (ALT) or USS, range from 2-3% in general paediatric populations, to between 10-30% in obese children or adolescents. These studies suggest that NAFLD is rare before adolescence, findings confirmed in a recent post-mortem study that provides the most robust evidence of the potential importance of NAFLD in adolescents. In that study, of 742 US individuals (aged 2-19 years) who had died from external causes (mostly road traffic accidents) the age, gender and ethnicity standardised prevalence of NAFLD (defined as greater than 5% steatosis on histology of the post-mortem liver) was 9.6%, with 3% of the population having NASH. The prevalence increased with increasing age from 0.7% in those ages 2-4 to 17.3% in those aged 15-19; it was very rare before the age of 10. Of importance, as duration of NAFLD in adults is associated with increased likelihood of progression to severe advanced disease, this means that it is likely that individuals with NAFLD in adolescence are at a very high risk of severe liver pathology in adulthood if fatty infiltration of their liver is not reversed (since by definition having the condition from adolescence into adulthood is a longer duration than fatty infiltration first appearing in adulthood). In one small (N=9) case study of clinically obese children who were referred because of persistently elevated ALT levels, those who complied with a hypocaloric diet were reported as having reductions in ALT levels that paralleled their weight loss. However, stronger epidemiological evidence is required to determine whether evidence of NAFLD in early life is reversible or whether changes in risk factors in early life can reverse evidence of hepatic fatty infiltration, liver damage and its associated adverse metabolic and vascular traits.

We recently found that 6% of a general (healthy) population of white American adolescents (age 12-19 years) had elevated ALT levels indicative of NAFLD (using a threshold of greater than =30 U/L to define NALFD). Furthermore, we showed marked differences in the distribution of ALT by gender (age standardised mean levels lower in girls compared with boys even after adjustment for body mass index (BMI)), which raises the question as to whether the same threshold of defining NAFLD in this age group should be used. We also found mean levels varied by ethnicity and that overtime mean ALT levels in adolescence have increased in this population. We used the publicly available US National Health and Nutrition Examination Survey (NHANES), which is an annually repeated cross-sectional survey that is used for public health surveillance and epidemiological studies in the US. Similar surveys - the Health Survey for England, the Scottish Health Survey and the Welsh Health Survey (http://www.natcen.ac.uk/natcen/pages/or_health.htm) - exist in the UK and are similarly used for public health surveillance and epidemiological studies. We sought to undertake a study of prevalence of elevated ALT and variation in ALT by gender, ethnicity and over time in adolescents in the UK but found that our national surveys did not have information on biomarkers for NAFLD. The lack of such data can be justified because, despite our and others attempts to examine the prevalence of NAFLD in adolescents using liver enzymes as biomarkers, there is currently insufficient evidence evaluating the utility of liver enzymes as biomarkers for NAFLD in adolescents in general or in the UK adolescent population in particular. Liver enzymes are relatively cheap and easy to assay. They would, therefore, be ideal for public health surveillance if they are shown to be accurate as diagnostic biomarkers in adolescents.

The diagnosis of NAFLD is suspected when there are indicative changes in liver enzymes (ALT, aspartate aminotransferase (AST), ALT:AST) and/or echogenic (fatty) liver detected by USS, together with the absence of any established causes of liver disease. A definitive diagnosis, and determination of severity, requires a liver biopsy, which is not feasible in large-scale epidemiological studies. Indeed, since liver biopsies are painful and have rare, but potentially severe, iatrogenic effects even in clinical practice liver biopsy is only recommended when there is a high level of suspicion of liver damage and clear indications that biopsy would alter clinical practice. The British Society of Gastroenterology and the British Association for the Study of the Liver guidelines for the use of liver biopsy in clinical practice state that the role of liver biopsy in the diagnosis and clinical management of NAFLD is currently not established. Similarly the American Gastroenterology Association have noted that the potential health risks, high cost and biopsy sampling errors associated with liver biopsy should be weighted against the value of the information that would be gained from performing a biopsy each patient. In clinical practice a combination of clinical history, measures of adiposity, presence of other associated morbidities, such as diabetes, elevated aminotransferases (ALT, AST and ALT:AST) and imaging modalities (USS, CT scan and MRI) are used and recommended in clinical guidelines for the diagnosis of NAFLD.

In research, elevated aminotransferases (ALT, AST, ALT:AST) have been predominantly used for case definition of NAFLD. However, there is marked variation in the thresholds used to define NAFLD across different studies, even when the same or very similar populations are used. For example, three studies have used ALT to assess NAFLD prevalence in adults from the USA NHANES survey and have used different thresholds - 19, 31 or 43 u/L in women; 40 or 43 u/L in men. Far fewer studies have examined the prevalence of NAFLD in general adolescent populations, but similar variation is found. ALT thresholds of between 30-40 u/L have been use, with identical thresholds used in females and males, despite clear evidence that in adolescence (as in adulthood) the distribution of ALT varies by gender.

There is biological plausibility for all three of ALT, AST and GGT as biomarkers for NAFLD; all are secreted by hepatocytes in response to liver damage, including NAFLD. ALT is the liver enzyme most commonly used in the assessment of liver conditions in general, including NAFLD, though its primary use has been questioned. It is a more specific marker of liver damage than AST, which is also found in the heart, skeletal muscle, kidney and other organs, and than GGT, which is expressed by most cells in humans. ALT elevation is greater than AST elevation in NAFLD and ALT is a more robust predictor of diabetes. Nonetheless elevation in both aminotransferases, ALT and AST, are recommended in clinical guidelines as biomarkers for NAFLD in adults. In NAFLD, plasma levels of both ALT and AST may be increased due to leakage from hepatocytes that have been damaged by fat accumulation. In addition, as ALT has an important role in gluconeogenesis and amino acid metabolism its greater elevation (compared to AST) may reflect increased gluconeogenesis, due to hepatic insulin resistance that is closely associated with liver fat accumulation, in addition to leakage from hepatocytes. Because of the greater ALT compared to AST elevation with NAFLD a low (below 1) AST:ALT ratio has also been suggested as a valid biomarker for NAFLD. The low AST:ALT ratio contrasts to the high (above 1) AST:ALT ratio that is typical of alcoholic liver disease.

GGT levels also increase with NAFLD, although this enzyme is not mentioned in clinical guidelines as part of the screening or diagnostic procedures for NAFLD. This is probably because it is has long been used as a marker of alcohol consumption and because it is expressed in most cell types and is therefore not a liver specific enzyme. In NAFLD, leakage from damaged hepatocytes would explain elevated levels of GGT. Thus elevated GGT levels could equally represent excess alcohol intake or NAFLD. Few adolescents are likely to have experienced the chronic alcohol abuse that is required to cause liver damage. Thus, it is biologically plausible that in adolescents GGT is also a useful biomarker for NAFLD. In this application we will use genetic variants associated with greater fat mass as proxies for fat related increase in GGT, and other enzymes (i.e. NAFLD related enzyme increase). These genetic variants are not associated with alcohol consumption and therefore we are able to use them to examine the extent to which elevated GGT, ALT and AST might be explained by NAFLD as opposed to alcohol consumption or other non-fat causes of liver damage.

Few studies have formally evaluated the utility of ALT, AST, AST:ALT and GGT as biomarkers for NAFLD in adolescents. One study in adults with the more severe form of NAFLD (non-alcoholic steatohepetitis (NASH)) examined the utility of ALT and AST as biomarkers for treatment effects in randomised controlled trials. The motivation of the investigators was their recognition of the need for relatively cheap and non-invasive surrogates even in RCTs of adults with more advanced disease (in whom biopsy would be more likely that in public health / epidemiological studies or trials of preventive interventions in health adolescents). Within a subgroup (N=102) in the trial they compared changes in ALT and AST levels to the main outcomes assessed by liver biopsy. They concluded that both aminotransferases were equally useful in RCTs of treatment for NASH in adults. Utility (assessed by receiver-operating characteristic (ROC) curves) was improved in multivariable analyses that adjusted for baseline ALT and AST levels as well as histological changes obtained from liver biopsy at baseline. In children and adolescents it has been suggested that elevated ALT levels potentially miss up to 40% of individuals with NAFLD.5;31These statements have been made largely on the basis of comparisons of elevated ALT with liver biopsy in clinical obese paediatric populations, who are unlikely to be representative of general populations. Even within clinically obese paediatric populations there are inconsistencies in the findings. For example, one study of obese Japanese children (up to age 16) found that a threshold of 30 u/L of ALT had a sensitivity of 0.92 (i.e. identified 92% of cases) for detecting NAFLD proven by USS. That study tested a particular threshold rather than using ROC curves33 to identify a threshold that maximises both sensitivity and specificity.

Thus, to date, there has been no thorough evaluation of ALT, AST, AST:ALT and GGT as biomarkers for NAFLD in general populations of adolescents. If we find that ALT, AST and GGT either alone or in combination are valid biomarkers for NAFLD they will provide a cheap and easily available marker that could be used widely for public health surveillance, epidemiological research and randomised controlled trials of preventive interventions.

In a healthy liver there should be no or very little fatty infiltration. The formal definition of NAFLD is that just 5% (or more) of hepatocytes have fatty infiltration on biopsy. On USS fat appears as a bright area. The USS indicators of fatty infiltration in the liver are bright hepatic echotexture (compared to the kidneys and/or spleen), deep attenuation and vascular blunting, and the extent of fatty infiltration can be quantified (none, mild, moderate and severe), with this quantification also shown to reflect amount of fatty infiltration on histology. In healthy individuals the liver should appear similar in echotexture to the kidneys and spleen, indicating no fatty infiltration; any indication of fatty infiltration supports a diagnosis of NAFLD if there are no other known causes. USS determined fatty liver has been shown in several studies to have high levels of sensitivity (89-90%), specificity (82-93%), positive predictive value (87-93%) and negative predictive value (87-94%), when compared with the gold standard of liver biopsy histology, for diagnosing moderate to profound levels of hepatic fatty infiltration. USS fatty infiltration has high levels of intra- and inter-rater reliability. USS cannot adequately identify liver fibrosis or cirrhosis in general asymptomatic individuals. However, neither can other radiological approaches. Indeed at present fibrosis and cirrhosis can only be diagnosed by biopsy, which as discussed above is not ethical or practical in any research study in health individuals.

Other radiological approaches including standard MRI, MR spectroscopy and CT scan are also able to accurately diagnose NAFLD. These modalities can identify different patterns of hepatic fatty infiltration, such as diffuse or nodular, but the relevance of these is currently unclear and the subject of on-going research. As yet there is no clear evidence that CT, MRI or MR spectroscopy have important benefits when compared to USS in simply identifying fatty liver infiltration.

Thus, USS provides a valid gold standard against which to evaluate ALT, AST, AST:ALT and GGT as biomarkers for NAFLD in adolescents.