Liver function tests: defining what's normalBMJ 2004; 328 doi: http://dx.doi.org/10.1136/bmj.38055.451968.55 (Published 22 April 2004) Cite this as: BMJ 2004;328:987
- Paul Roderick (), senior lecturer in public health
- Community Clinical Sciences Research Division, University of Southampton, B Floor, South Academic Block, Southampton General Hospital, Southampton SO16 6YD
- Correspondence to: P Roderick
Chronic liver disease and hepatocellular carcinoma are major worldwide public health problems in countries with endemically high levels of viral hepatitis (B and C).1 However, even in western countries chronic liver disease is an emerging problem, due not only to viral hepatitis but also to the effects of lifestyle factors such as heavy alcohol consumption and obesity.2 3
Liver function tests are widely performed blood tests used in patients with suspected liver disease or unexplained illness and in some specific situations such as screening of blood donors. The most widely used tests are those used to detect the aminotransferases—alanine and aspartate—which are associated with hepatocellular injury. Raised concentrations may indicate serious underlying chronic liver disease, recognition of which is important for guiding interventions to modify lifestyle and use of specific therapies such as interferon for hepatitis C to prevent the risk of progression to cirrhosis.
The sensitivity, specificity, and predictive values are important in assessing the clinical utility of such tests. Normal ranges have been based on distributions from healthy volunteers with two SD above the mean (that is, top 2.5% cut-off) being considered the upper normal range. Aminotransferase concentrations maybe within the normal range in people with chronic liver disease.4 5 There is ongoing debate about whether to lower the normal range to take account of changing lifestyle factors that influence aminotransferase concentrations, particularly obesity, which in Western countries would increase the detection of hepatitis C and alcoholic and non-alcoholic fatty liver disease (NAFLD).6 7
Kim and colleagues have analysed the association between aminotransferase concentrations and mortality from liver disease in a large prospective cohort in Korea.1 They found that there was a graded increase in risk of mortality from liver disease even within the normal range (20-40 IU/l) compared with the lowestconcentrations (< 20 IU/l) for both sexes. The performance of the test in identifying future risk of mortality from liver disease was maximised by lowering the threshold to about 30 IU/l. They propose identifying a borderline level of aminotransferase of 30-39 IU/l, suggesting that patients in this category should be further investigated with more specific diagnostic tests for chronic liver disease.
What are the implications for countries with lower levels of chronic liver disease than Korea? Test utility is affected by disease prevalence. In countries with lower prevalence the negative predictive value (that is, the ability of a negative test to exclude disease) will be higher and the positive predictive value (that is, the ability of a positive test to predict disease) will be lower. Lowering the upper normal range for aminotransferase or including borderline cases will increase sensitivity at the expense of specificity, so detecting more cases of chronic liver disease but with a lower positive predictive value. These effects would be less marked in those with suspected liver disease compared with the population as whole. The cost effectiveness of further investigation of borderline cases is not known but requires evaluation. A crucial diagnostic issue is how to identify severe chronic liver disease (indicated by inflammation and fibrosis), which is associated with high risk of progressing to cirrhosis. Definitive evaluation currently relies on liver biopsy. Research is needed to evaluate combinations of non-invasive measures to predict severe liver disease; this could include patients with borderline concentrations of aminotransferase.
Competing interests Research collaboration with Bayer Health-Care.