Value of determining carbohydrate-deficient transferrin isoforms in the diagnosis of alcoholic liver disease

OBJECTIVE: To determine whether isoform separation of carbohydrate-deficient transferrin (CDT) is of value in the diagnosis of alcoholic liver disease (ALD) and is specific to ALD when compared with other liver diseases. PATIENTS AND METHODS: During 1995 and 1996, 47 patients with ALD were evaluated with CDT at the Mayo Clinic in Rochester, Minn. The diagnosis of ALD was based on biochemical and histological analyses and on a history of drinking that exceeded 5 years with an average alcohol intake of more than 60 g/d. Disease controls included nonalcoholic steatohepatitis (NASH) (n = 26) and other liver disease (n = 22). Normal controls (n = 21) were healthy individuals without liver disease. Transferrin isoforms were quantified by densitometry of Coomassie-stained transferrins after affinity purification and isoelectric focusing. The pentasialo, tetrasialo, trisialo, disialo, monosialo, and asialo isoforms were quantified as percentages of total band densities. RESULTS: Receiver operating characteristic (ROC) curves were constructed for each isoform. The curves for total desialated isoforms (sum of disialo, monosialo, and asialo) displayed the best relationship between sensitivity and specificity with an ROC-area under the curve (AUC) of 0.922. The ROC-AUC values for individual transferrin isoforms in ALD vs NASH for pentasialo, tetrasialo, trisialo, disialo, monosialo, and asialo were 0.806, 0.917, 0.885, 0.933, 0.804, and 0.785, respectively. Only 58% of patients with ALD were detected at a specificity that excluded ALD in 84% of those who did not have it. CONCLUSION: Within alcohol ingestion times reported to us, no associations with recent drinking were observed. Alcohol as a cause of liver disease is not perfectly established by CDT analysis, although a high total CDT value favors ALD over NASH.