Background and aim: The incidence of hepatocellular carcinoma is high among cirrhotic patients, ranging from 2 to 3% in western cohorts and 6-11% in eastern cohorts. Although only one randomised trial has been performed, clinical practice generally uses periodic screening to detect hepatocellular carcinoma in cirrhotic patients. We reviewed the scientific literature on hepatocellular carcinoma screening.
Materials and methods: Evaluation of studies identified through MEDLINE and EMBASE (1990-May 2003).
Results: The available screening tests to detect hepatocellular carcinoma are alpha-fetoprotein (cut-off: 20 ng/ml) and ultrasound, which are generally combined. The reported sensitivity and specificity are 50-85% and 70-90%, respectively. An estimated doubling time of about 6 months has led to the use of an interval of 6 months between screenings. Based on the risk of hepatocellular carcinoma, cirrhotic patients are considered as the target population. Screening seems to detect smaller and more frequently unifocal hepatocellular carcinoma; the residual liver function is important for determining the eligibility for effective treatment (resection); hence the prevention is more effective for patients with well-compensated cirrhosis. The survival estimated by non-randomised studies is slightly longer for patients with screening-detected hepatocellular carcinoma, compared to those with clinically detected hepatocellular carcinoma, although few studies have accounted for 'lead time bias'.
Conclusions: Although screening for the early detection of hepatocellular carcinoma has become quite common in clinical practice, its effectiveness remains controversial. Observational studies that have taken into account lead time bias suggest that survival is greater for patients with screening-detected hepatocellular carcinoma, yet the eligibility for effective treatments is low. Considering that only one randomised controlled trial has been conducted, it is crucial to standardise the screening schedule and to evaluate prevention programmes.