Is inconsistency of α-fetoprotein level a good prognosticator for hepatocellular carcinoma recurrence?

AIM: To identify the clinical outcomes of hepatocellular carcinoma (HCC) patients with inconsistent a-fetoprotein (AFP) levels which were initially high and then low at recurrence. METHODS: We retrospectively included 178 patients who underwent liver resection with high preoperative AFP levels (>= 200 ng/dL). Sixty-nine HCC patients had recurrence during follow-up and were grouped by their AFP levels at recurrence: group I, AFP <= 20 ng/dL (n = 16); group II, AFP 20-200 ng/dL (n = 24); and group III, AFP >= 200 ng/dL (n = 29). Their preoperative clinical characteristics, accumulated recurrence rate, and recurrence-to-death survival rate were compared. Three patients, one in each group, underwent liver resection twice for primary and recurrent HCC. AFP immunohistochemistry of primary and recurrent HCC specimens were examined. RESULTS: In this study, 23% of patients demonstrated normal AFP levels at HCC recurrence. The AFP levels in these patients were initially high. There were no significant differences in clinical characteristics between the three groups except for the mean recurrence interval (21.8 +/- 14.6, 12.3 +/- 7.7, 8.3 +/- 6.6 mo, respectively, P < 0.001) and survival time (40.2 +/- 19.9, 36.1 +/- 22.4, 21.9 +/- 22.0 mo, respectively, P = 0.013). Tumor size > 5 cm, total bilirubin > 1.2 mg/dL, vessel invasion, Child classification B, group Ill, and recurrence interval < 12 mo, were risk factors for survival rate. Cox regression analysis was performed and vessel invasion, group III, and recurrence interval were independent risk factors. The recurrence interval was significant longer in group I (P < 0.001). The recurrence-to-death survival rate was significantly better in group II (P = 0.016). AFP staining was strong in the primary HCC specimens and was reduced at recurrence in group I specimens. CONCLUSION: Patients in group I with inconsistent AFP levels had a longer recurrence interval and worse recurrence-to-death survival rate than those in group II. This clinical presentation may be caused by a delay in the detection of HCC recurrence. (C) 2010 Baishideng. All rights reserved.