We read with interest the updated hepatocellular carcinoma (HCC) guidelines by the American Association for the Study of Liver Diseases.1 We were surprised by the omission of alpha-fetoprotein (AFP) testing in the recommendations for HCC surveillance. We disagree with these recommendations.
In making recommendations, the writers of practice guidelines should consider the quality of the evidence. The HCC guidelines ignore a significant amount of data about the use of AFP in the surveillance of patients at risk for HCC. The only available level 1 evidence for HCC surveillance comes from one randomized controlled trial of ultrasonography (US) combined with AFP testing every 6 months in a hepatitis B carrier population.2 The next best available evidence comes from a population-based cohort surveillance program involving hepatitis B carriers in Alaska that showed improved outcomes.3 The remainder of the literature includes population-based and non–population-based cohorts and case-control studies open to multiple sources of bias.4, 5 Although it may be reasonable to generalize the findings of the available randomized trial and population-based study to other patient groups with cirrhosis or hepatitis C, we feel that it is inappropriate to drop one of the interventions (i.e., AFP) found to work.
The guidelines cite the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) study as the main source for the lack of efficacy of AFP in patients with cirrhosis.6 There are significant limitations to this study. First, only 40% of the patients had cirrhosis. Second, HCC surveillance was not the primary purpose of HALT-C. Third, AFP had a sensitivity and specificity at the time of HCC diagnosis of 61% and 81%, respectively, whereas US had a sensitivity of only 58%, which is inadequate according to the criteria stated in the guidelines. Interestingly, 40% of the patients with early-stage HCC were diagnosed by an increasing AFP level alone or in combination with US. Therefore, AFP appears to complement US for the surveillance of HCC.
In addition to ignoring the highest level of evidence for the efficacy of US combined with AFP in research studies, the HCC guidelines also neglect the effectiveness of the tests in clinical practice. Test reproducibility, a major determinant of translating the results of research studies into practice, has never been evaluated for US as an HCC surveillance test. Another issue is underutilization of surveillance tests. In the only population-based study evaluating surveillance for HCC, only 17% of patients with HCC underwent regular surveillance before their diagnosis.7 Dropping AFP from the guidelines may potentially lower the percentage of patients undergoing surveillance.
Surveillance for HCC has a whole host of confounding factors that make it impossible to detect benefit through personal experiences and clinical observations alone.8 Therefore, randomized controlled studies are the only reliable way of evaluating surveillance and changing clinical practice. In the absence of randomized studies in patients with cirrhosis, the current evidence points to US combined with serum AFP as the most effective surveillance strategy for patients at risk for HCC. The guidelines should be revised to recommend US with AFP as the best available surveillance strategy.