RESUMO
BACKGROUND: Hepatocellular carcinoma is the most common cause of primary liver neoplasms and is one of the main causes of death in patients with liver cirrhosis. High Alpha fetoprotein serum levels have been found in 60-70% of patients with Hepatocellular carcinoma; nevertheless, there are other causes that increase this protein. Alpha fetoprotein levels > or =200 and 400 ng/mL in patients with an identifiable liver mass by imaging techniques are diagnostic of hepatocellular carcinoma with high specificity. METHODS: We analysed the sensitivity and specificity of the progressive increase of the levels of alpha fetoprotein for the detection of hepatocellular carcinoma in patients with liver cirrhosis. Seventy-four patients with cirrhosis without hepatocellular carcinoma and 193 with hepatic lesions diagnosed by biopsy and shown by image scans were included. Sensitivity and specificity of transversal determination of alpha fetoprotein > or = 200 and 400 ng/mL and monthly progressive elevation of alpha fetoprotein were analysed. Areas under the ROC curves were compared. Positive and negative predictive values adjusted to a 5 and 10% prevalence were calculated. RESULTS: For an elevation of alpha fetoprotein > or= 200 and 400 ng/mL the specificity is of 100% in both cases, with a sensitivity of 36.3 and 20.2%, respectively. For an alpha fetoprotein elevation rate > or =7 ng/mL/month, sensitivity was of 71.4% and specificity of 100%. The area under the ROC curve of the progressive elevation was significantly greater than that of the transversal determination of alpha fetoprotein. The positive and negative predictive values modified to a 10% prevalence are of: 98.8% and 96.92%, respectively; while for a prevalence of 5% they were of 97.4% and 98.52%, respectively. CONCLUSION: The progressive elevation of alpha fetoprotein > or =7 ng/mL/month in patients with liver cirrhosis is useful for the diagnosis of hepatocellular carcinoma in patients that do not reach alphaFP levels > or =200 ng/mL. Prospective studies are required to confirm this observation.
