Diagnosis of hepatocellular carcinoma in patients with liver cirrhosis using liver function assays

Sex, age and 21 routine liver function assays were analyzed by stepwise selection and the best-of-all-possible-combinations method to identify a small group of assays valuable in establishing which liver cirrhosis (LC) patients have a high risk of hepatocellular carcinoma (HCC), when alpha-fetoprotein (AFP) is not elevated. Data was obtained from 115 HCC and 122 LC patients on admission. Tumor size correlated with AFP (0.73), alkaline phosphatase (ALP, 0.47), leucine aminopeptidase (LAP, 0.42), lactic dehydrogenase (LDH, 0.42), and the glutamic oxaloacetic transaminase (GOT)/glutamic pyruvic transaminase (GPT) ratio (GOT/GPT, 0.41). The mean of the correct diagnosis rates (CDR) of HCC and LC utilizing AFP as the sole parameter (89%) was markedly higher than those of the other parameters. The best-of-all-possible-combinations method presented a more powerful combination than stepwise selection. The best combination of 7 parameters (LAP, GOT/GPT, choline esterase, one-hour erythrocyte sedimentation rate, age, albumin/globulin ratio, and total bilirubin) presented a mean CDR of 80%, HCC CDR of 77%, and false positive rate of 18%. LC patients statistically diagnosed as having HCC by these 7 parameters are proposed as high risk patients. Fourteen (78%) of 18 HCC patients who were AFP-negative were statistically diagnosed. This analysis can be applied to LC patients to distinguish those that should be followed closely by imaging diagnostic techniques.