Comparison of 2 human chorionic gonadotropin assays as tumor markers assays.

Intact human chorionic gonadotropin (hCG), hyperglycosylated hCG, and the free subunit are the forms of hCG produced by healthy tissues and tumors (1 ). Intact hCG is the predominant form in serum and plasma in pregnancy and trophoblastic disease, hyperglycosylated hCG in choriocarcinoma and testicular cancer, and the free subunit in nongestational malignancy. Proteases present in macrophages associated with tumors or present in serum degrade these forms, resulting initially in nicked intact hCG, nicked hyperglycosylated hCG, and nicked free subunit. Nicked intact hCG rapidly dissociates into free nicked subunit and free subunit. The C-terminal peptide is cleaved, and further degradation results in the -subunit core fragment (1 ). All of these variants may be present in serum and urine, and an ideal hCG tumor marker assay would detect all of these products. Tumors that produce only free subunits can be most effectively detected by assays that specifically measure this form (2 ). Detection of -subunit core fragment in serum or plasma may not be important because concentrations are exceedingly low (3 ). In a recent review, the authors reported that the Siemens Immulite hCG assay is the only assay that is useful for all pregnancy, gestational trophoblastic disease, and cancer applications (1 ). However, this conclusion is not supported by data from 2 recent studies in which WHO International Reference Reagents were used for all the major variant forms of hCG. These 2 studies revealed molar crossreactivities for the Roche hCG (also known as Roche Total hCG) assay as follows: nicked hCG (96% and 95.8%,), free (130% and 107.7%), nicked free (92% and 92.5%), and -core fragment (33% and 38.6% ). Although recognition of the -core fragment by the Roche assay was somewhat lower than by the Immulite (53%) and Immulite 2000 (63% and 68.3%), this finding is of importance only if the assay is to be used for measurements of hCG concentrations in urine (4, 5 ). We analyzed 191 patient serum samples submitted to 2 New Zealand diagnostic laboratories for tumor-marker hCG measurement using both the Roche hCG and the Immulite 2000 assays. Clinical information was available for 96 of the samples: 46 from patients with gestational trophoblastic disease, of which 76% were positive ( 5 IU/L); 35 with germ-cell tumors (80% positive); 8 with unspecified gonadal tumors (50% positive); and 7 with other cancers (71% positive). hCG concentrations in the samples varied from 1 to 373 000 IU/L, with the majority of results 1000 IU/L. The upper limit of the reference interval for serum hCG in our laboratories is 5 IU/L. Discrepant results were defined as those that were either 7 IU/L on Immulite and 4 IU/L on Roche assay, or 4 IU/L on Immulite and 7 IU/L on Roche assay; this definition was used in order not to classify as discrepant closely similar results that happened to lie on either side of the cutoff point. Among the 191 samples tested there was only 1 discrepant result (7.4 IU/L on Immulite and 3.4 IU/L on Roche assay). The relationship between the 2 assays is shown in Fig. 1. Weighted Deming-regression analysis of results 100 000 IU/L yielded the following equation: Roche hCG (0.85 Immulite hCG) 0.57. Weighted Deming-regression analysis with a constant CV was performed by using Microsoft Excel with Analyze-it software. In 1 patient with a hydatidiform mole, the initial hCG result measured on the Immulite was falsely low, owing to antigen excess