Validity of Alpha Fetoprotein for Diagnosis of Hepatocellular Carcinoma in Cirrhosis

Objective: To determine the accuracy of serum alpha fetoprotein (AFP) for diagnosis of hepatocellular carcinoma (HCC) in patients with cirrhosis. Study Design: Observational study. Place and Duration of Study: Department of Medicine, The King Edward Medical University, Lahore, from November 2007 to August 2011. Methodology: Consecutive patients, diagnosed with HCC by contrast enhanced CT, MRI or biopsy were included as cases. Patients of cirrhosis with no evidence of HCC were enrolled as controls. Demographic, laboratory and radiological data were recorded. Serum AFP was determined in all patients at outset and was analyzed using ROC curve for its accuracy in diagnosing HCC. Results: A total of 275 patients were included; of them 173 had HCC and 102 had cirrhosis. One hundred and thirty nine cases (80.3%) with HCC and 86 (84.3%) without HCC had cirrhosis due to HCV. Stage of liver disease, as determined by Child Turcotte Pugh (CTP) score, was comparable; mean CTP value 7.97 +/- 2.17 in HCC and 7.75 +/- 2.21 in control group (p = 0.41). Area under curve (AUC) for AFP was 0.85 (95%, CI: 0.80 - 0.90) with optimum cut off value of 20.85 ng/ml which showed 72.2% sensitivity, 86.2% specificity, 89.9% positive predictive value, 64.7% negative predictive value and 77.4% overall accuracy in diagnosing patients with HCC. Conclusion: Despite sub-optimal sensitivity, alpha fetoprotein is still a valid screening test for diagnosis of hepatocellular carcinoma till other more sensitive markers are developed. Till then, it should be used in conjunction with ultrasound.