Standardization of assays for human chorionic gonadotropin.

The unsatisfactory standardization of many immunoassays is fairly well recognized, but the problem is hard to tackle (1 ). Recently, studies on the performance of immunoassays for testosterone in female serum (2 ) and for urinary cortisol (1 ) have revealed that the quality of some widely used assays is totally unsatisfactory. The quality of assays for human chorionic gonadotropin (hCG) has also been shown to be problematic (3, 4), and this was a reason for the IFCC to establish a working group for standardization of assays for this hormone. In addition to being a clinically important analyte, hCG immunoreactivity in biological fluids comprises a reasonable number of wellknown variants. hCG was thus considered a suitable model for standardization of glycoprotein hormones in general (5 ). It has been argued that it is impossible to standardize assays for heterogeneous antigens (6 ), and an important aim of the IFCC committee on hCG was to study whether this is the case. Other aims were to introduce a uniform nomenclature, to prepare new standards, and to assign values to these in substance concentrations (mol/L). Standards should preferably be pure, identical to the analyte in serum, and homogeneous, but glycoproteins are inherently heterogeneous because of variation in the carbohydrate moieties. Standards for subunits and degraded forms of hCG occurring in circulation were also needed (5 ). The working group of the IFCC has achieved most of its goals. Standards for hCG and its most important variants, i.e., its and -subunits (hCG and hCG ), the core fragment of hCG (hCG cf), and two partially cleaved or “nicked” forms (hCGn and hCG n), are now available from WHO as the 1st Research Reagents (1st RRs) (7 ). The hCG preparation was used by Cole et al. (8 ) to study how nine widely used immunoassays for hCG are standardized. In addition, the reactivity of these assays with several hCG variants was examined. Their study provides interesting information on causes of between-assay variation and some means to cure these problems. When serum samples with additions of various partially purified hCG preparations were analyzed with the nine assays, the difference between the highest and lowest result was 1.4to 1.6-fold. If the results were recalculated on the basis of those obtained for the new 1st RRs for hCG, the maximum variation was reduced to 1.27-fold. The authors draw the conclusion that use of impure assay calibrators in combination with variable recognition of hCG subunits and other variants is an important cause of the variation. Analysis of several commercial hCG preparations revealed that most of them contain degraded forms, subunits, and fragments (8 ), and earlier studies have shown that rather crude hCG preparations are used as calibrators (4 ). The recognition of hCG variants by the studied immunoassays was analyzed by use of purified preparations of hCG , hCGn, hCG cf, and hyperglycosylated hCG. Recognition of these was much more variable than that of “normal” hCG. The immunoreactivities of the variants based on molar concentrations can be estimated on the basis of their molecular weights (7 ) and the ratio between content in mass units and IU of the 3rd International Standard (IS); i.e., 1 g 9.28 IU 26.7 pmol (5 ). When analyzed by this method the mean value of hyperglycosylated hCG (900 IU/L 97 g/L 2590 pmol/L) is fairly close to the nominal value (84 g/L), although one assay underestimated and another one overestimated this variant (52% and 172% of mean, respectively). Thus, hyperglycosylated and normal hCG are recognized fairly equally by most of the assays. This is also the case with hCGn; the mean value (3460 IU/L) corresponds to 373 g/L, whereas the nominal value is 330 g/L. The measured mean concentration of hCG (3976 IU/L) corresponds to 11 440 pmol/L hCG, which is equal to 269 g/L hCG , whereas the nominal concentration was 230 g/L. (The values for hCG are reported in IU for hCG, according to the 3rd IS, not in those for the 3rd IS for hCG , which would give 9.28-fold lower values.) Only one assay detected hCG cf. The concentration measured, 130 IU/L, corresponds to 0.374 nmol/L hCG, and this corresponds to 3.74 g/L hCG cf, i.e., similar to the nominal concentration, 4 g/L. Thus, with the exception of hCG cf, no variant is grossly overor underestimated by any of the assays studied (8 ). This comparison also demonstrates the advantage of using molar concentrations to express values for various forms of hCG (5 ). Cole et al. (8 ) also studied the effect of differences in sialic acid content, which cause extensive charge heterogeneity in hCG. Charge variants with pI values between 3 and 7 were equally recognized by all assays, showing that variation in sialic acid content does not affect immunoreactivity. This is in agreement with the results of another recent study and show that carbohydrate variation does not usually affect immunoreactivity (9 ). However, the carbohydrate moieties of hyperglycosylated hCG affected recognition in two assays (10 ), and one antibody recognizing only a certain form of hyperglycosylated hCG has been described. This antibody, B152, recognizes an epitope comprising the type 2 o-core carbohydrate on Ser132 together with part of the C-terminal peptide of hCG . This antibody does not recognize normal hCG containing type 1 o-core carbohydrates (10 ). The epitope specificities of several antibodies used by assay manufacturers have been carefully characterized (11 ), but it is not known which antibodies are actually used in the assays studied. It would be interesting to know which epitopes are recognized by the antibodies in assays with aberrant recognition of hyperglycosylated hCG. What is the significance of the between-assay variation observed? It is obvious that a misleading result may be obtained if different assays are used during monitoring of a patient with suspected ectopic pregnancy or threatening abortion during early pregnancy, when most of the hCG Editorial