Glycan composition of serum alpha-fetoprotein in patients with hepatocellular carcinoma and non-seminomatous germ cell tumour.

Although estimation of serum alpha-fetoprotein (AFP) is widely used in the diagnosis of hepatocellular carcinoma (HCC) and non-seminomatous germ cell tumours (NSGCT), the clinical usefulness of this test is limited by a low specificity. However, there exist glycoforms of AFP which may be more specific for particular tumours. Previously, detailed analysis has been prevented by the low levels of AFP in human serum. We report here the application of fluorescence labelling, sequential exoglycosidase digestion, high-performance liquid chromatography and matrix-assisted laser desorption ionization in time-of-flight mass spectrometry, to determine the glycan structures of purified serum AFP from patients with HCC and NSGCT. Eleven major glycans were found, of which seven were N-linked, and four were O-linked, to the protein backbone. The structure of the N-linked glycans (all of bi-antennary complex-type with varying degrees of sialylation, fucosylation and galactosylation) were consistent with those previously reported. The O-linked glycans (three mucin O-GalNAc type glycans with variable degrees of sialylation, one O-HexNAc monosaccharide glycan) have not previously been reported. The finding of mucin O-GalNAc type glycans was supported by the prediction of potential O-GalNAc glycosylation sites on the protein backbone by analysis of the AFP structure by molecular modelling. With knowledge of these structures it may be possible to develop more specific assays for the detection of HCC and NSGCT.

'Hepatoma-specific' alphafetoprotein may permit preclinical diagnosis of malignant change in patients with chronic liver disease.

The only hope for effective treatment of hepatocellular carcinoma (HCC or 'hepatoma') lies in early diagnosis. Measurement of the serum alphafetoprotein (AFP) level is potentially a useful screening test. When grossly raised, it is almost diagnostic of HCC. However, modestly elevated levels may also arise in patients with benign chronic liver disease, and this markedly decreases the test's specificity and hence its clinical value. In 582 consecutive attendees at an outpatient clinic for people with chronic liver disease, a single blood sample was taken for analysis of 'total' AFP and the 'hepatoma-specific' AFP isoform. Using ultrasonography as the primary screening method, patients with AFP levels > or = 50 ng ml-1 were followed up throughout the study or until HCC was diagnosed on the basis of conventionally defined criteria. On entry into the study, 53 patients had an AFP concentration > = or 50 ng ml-1 and the 'hepatoma-specific' AFP isoform was detected in 26 of these. During an 18-month follow-up period, a diagnosis of HCC was established by conventional methods in 19 (17 'definite' and two 'probable') of these 26 patients. In only two cases was there ultrasound evidence of tumour development at the time AFP was first found to be elevated; in the remainder a diagnosis of HCC, based on ultrasound screening, was established at a median time of 3.6 months (range 1-18 months) after entry into the study. Among those 27 without the 'hepatoma-specific' isoform, one developed a 'definite' HCC and two developed 'probable' tumours. With the application of 'hepatoma-specific' AFP, the positive predictive value of the test was 73.1%, compared with only 41.5% using the conventional 'total' AFP test. Application of this test for the 'hepatoma-specific' AFP markedly increases the positive predictive value of AFP and, in some cases, permits the presence of tumour to be inferred before it could be detected by routine ultrasound examination.