Abstract
Background
Alpha-fetoprotein (AFP) levels for the diagnosis of hepatocellular carcinoma (HCC) may vary by geographical region and racial background. No data exists for this test in the Middle Eastern population. In addition, there is limited data on the impact of virological status on AFP levels.
Methods
In a multicenter, case-control study involving 206 cases, 199 cirrhotic and 197 chronic hepatitis controls, we assessed the utility of AFP in the diagnosis of HCC (sensitivity, specificity, positive (PPV) and negative (NPV) predictive values, and positive likelihood ratios (LR). PPV and NPV were evaluated for three additional HCC prevalence rates (5, 10, and 20%).
Results
The best discriminating AFP value was 11.7 ng/ml. The sensitivity ranged from 32 to 79.5% at different AFP levels with the specificity increasing sequentially from 47.7 to 98.5%. Sensitivity of AFP at the best cut-off level for hepatitis C virus (HCV), hepatitis B virus (HBV) and non-viral etiology for HCC was 73.7, 65.6, and 59.5%, respectively. Specificity at this level for HCV, HBV, and non-viral etiology was 36.6, 30.1, and 29.4%, respectively. AFP cut-off levels of 102, 200, and 400 ng/ml showed similar sensitivity (39.8, 35.9, and 32%, respectively) and specificity (96, 98.5, and 98.5% respectively). Positive LR for AFP at >11.7, >20, >102, >200, >400 ng/ml were 2.8, 3.3, 9.9, 23.8, and 21.2, respectively.
Conclusions
In cirrhotic patients, AFP has a poor screening and diagnostic value for HCC. Underlying viral etiology fails to influence the diagnostic accuracy of this test. An AFP level greater than 100 ng/ml has a high degree of specificity and may be used as a confirmatory test.