Determination of hyperglycosylated human chorionic gonadotropin produced by malignant gestational trophoblastic neoplasias and male germ cell tumors using a lectin-based immunoassay and surface plasmon resonance

Abstract

The ability to reliably detect aberrant glycosylation of human chorionic gonadotropin (hCG) may have profound implications for the diagnosis and monitoring of malignant gestational trophoblastic neoplasia, germ cell tumors, other malignancies, and pregnancy complications. To become a clinically useful assay, however, this discrimination of glycoforms should be possible on minimally treated biological specimens. Towards this end, we have developed a lectin-based sandwich-type immunoassay to compare the glycosylation patterns of hCG among urine specimens from patients presenting with a normal pregnancy, invasive mole, choriocarcinoma, and male germ cell tumors using carbohydrate-free antibody fragments as capture reagents and a panel of eight lectins, five recognizing neutral sugars and three recognizing sialic acid. There was no significant difference in the binding of any of the lectins to hCG in the urine of women over the gestational range of 6-38 weeks. Three lectins, however, exhibited differential binding to urinary hCG derived from these normal pregnant controls and that from patients with malignant forms of gestational trophoblastic disease and male germ cell tumors. Galanthus nivalis agglutinin and Maackia amurensis lectin, which bind terminal mannose and alpha(2-3)sialic acid, respectively, preferentially bound pregnancy-derived hCG, whereas the lectin, wheat germ agglutinin, which binds sialic acid and beta(1-4)N-acetylglucosamine, exhibited decreased binding to pregnancy-derived hCG compared to that from patients with male germ cell tumors and malignant gestational trophoblastic neoplasia. The differential binding observed with these three promising lectins is most encouraging and warrants further examination. The experimental paradigm also holds promise for the development of comparable assays for other glycosylated tumor markers.

Figure 1

Immunoreactive forms of hCGβ in urine samples as determined by Western analysis. Samples EB, 16, 21, and 952–957 had been previously concentrated; the others are raw urine samples. Diluted samples were probed with monoclonal antibody B108, which recognizes an epitope common to hCG, hCGβ, and the hCGβ-core fragment, and serves as the capture antibody for the biosensor assay. A. Pregnancy samples are designated P#, with the number indicating the week of pregnancy at time of sampling. The three non-pregnant controls are designated NPF01-NPF03 B. Pregnancy (EB = 11 weeks and P14 =14 weeks) and patient disease samples. 16, 21, 954 = choriocarcinoma; 952, 955, 956 = invasive mole; 953, 957 = male germ cell tumor

Figure 2

Representative sensorgrams of lectin binding to hCG captured from urine specimens. Samples were diluted to equivalent amounts (150 ± 5 RU) of hormone so that their binding of lectins could be directly compared. The binding of GNA to hCG captured from one sample of each patient condition is shown.

Figure 3

Binding of lectins (mean ± SD) recognizing neutral sugars to 150 RU hCG captured from human urine samples, grouped according to condition and compared to non-pregnant controls with ANOVA. A. AAL. B. GNA. C. PHA-E. D. PSA. E. RCA-I. * = p< 0.05, ** = p<0.01

Figure 4

Binding of lectins (mean ± SD) recognizing sialic acid to 150 RU hCG captured from human urine samples, grouped according to condition and compared to nonpregnant controls with ANOVA. A. MAL II. B. SNA. C. WGA. * = p< 0.05, ** = p<0.01