Background: Radiographic measurements do not always reflect the biological response of hepatocellular carcinoma to drug therapy. Aims: To evaluate the clinical implications of tumour marker (alpha-fetoprotein) response in advanced hepatocellular carcinoma patients with thalidomide treatment. Patients and methods: Forty-two advanced hepatocellular carcinoma patients with baseline alpha-fetoprotein levels above 200 ng/mL and thalidomide therapy were included. Serum alpha-fetoprotein levels were measured every 4 weeks. alpha-fetoprotein response was defined as a 50% or greater reduction of alpha-fetoprotein levels for 4 or more weeks during treatment. Radiographic response was assessed by World Health Organization criteria; survivals were estimated by Kaplan-Meier method and prognostic factors were assessed by Cox's proportional hazard model. Results: With intention-to-treat analysis, radiographic response and alpha-fetoprotein response were obtained in 7% (three of 42, 95% confidence interval: 0-15) and 24% (10 of 42, 95% CI: 10-38) of patients, respectively. All radiographic response was observed in alpha-fetoprotein responders. Multivariate analyses showed alpha-fetoprotein response was independent prognostic factor for both progression-free survival (relative risk = 0.394, 95% CI: 0.189-0.820, P = 0.013) and overall survival (relative risk = 0.241, 95% CI: 0.096-0.606, P =0.003), whereas radiographic response was not. Conclusion: alpha-fetoprotein response can more accurately reflect the biological response of advanced hepatocellular carcinoma to thalidomide therapy than radiographic response.
