I read with interest the article by Asahina et al. regarding the levels of α-fetoprotein (AFP) and alanine aminotransferase (ALT) after interferon therapy that could predict the occurrence of hepatocellular carcinoma (HCC) in patients with hepatitis C virus (HCV) infection. The authors made a great effort to evaluate the levels of AFP and ALT in a cohort of 1,818 patients after interferon therapy and found that cutoff values for ALT and AFP for prediction HCC development were 40 IU/L and 6.0 ng/mL, respectively. These findings may be helpful for clinicians to closely follow-up high-risk patients to detect early-stage HCC. However, these data may also be misleading and need further clarification.
The levels of AFP and ALT were measured every 1-6 months. It is unknown which timepoint of AFP and ALT levels were selected for calculation. Serum AFP levels usually fluctuate during serial observation. Some patients may present with an AFP >6 ng/mL before therapy, which decreased to <6 ng/mL within a few months after interferon therapy, and became elevated to >6 ng/mL at long-term follow-up. Should these patients be classified as AFP ≧6 ng/mL decreased group or AFP ≧6 ng/mL unchanged group? The mean follow-up period of this study was 6.1 years. The follow-up period in this study has been as long as 20 years. If the interval of AFP measurement was as short as 1 month, too many unnecessary measurements could have been performed. As shown in the results, postinterferon therapy AFP level ≧6.0 ng/mL had a positive predictive value of only 0.262. Consistent with previous observation, this may evoke inappropriate suspicion of malignancy in 74 out of 100 patients with AFP above this cutoff value. A total of 179 patients developed HCC, accounting for 9.8% of the entire cohort of 1,818 patients. Based on the HALT-C trial, patients with persistent elevation of AFP after interferon therapy, only 2% were noted to develop HCC. We believe that a large proportion of patients still did not have elevated AFP levels on detection of HCC occurrence. Thus, the applicability and cost-effectiveness of this policy merits further investigation.