ABSTRACT
In 39 samples of pregnancy urine, the concentration of human chorionic gonadotrophin (HCG) was estimated biologically by the ovarian hyperaemia reaction in rats, and serologically by the passive haemagglutinationinhibition technique. The results of the bioassays varied from 3 to 150 IU/ml, those of the immunoassays from 5 to 640 IU-eq./ml, and the correlation between the two (calculated for their logarithms) accounted for only 17 per cent of the total variation (r2 = 0.169, P ≈0.01). If the biological activity were estimated from a serological result and the appropriate regression line, the fiducial interval for P = 0.05 would extend from 17 to 610 per cent of the estimate.
In a factorial experiment using three anti-HCG sera, three standard and three sensitizing preparations of HCG, the sensitivity of the serological system (expressed as the endpoint concentration in IU of HCG) varied considerably between the 27 combinations of the 3 factors, but there was no interaction between the latter.
From these data and those of other authors, it is concluded that immunoassays based on haemagglutination inhibition cannot replace bioassays in the estimation of HCG, as distinct from its hypothetical metabolites or other related antigens, unless specificity has been demonstrated. The well-documented reliability of serological pregnancy tests is, of course, not in dispute.
Attention is drawn to the fact that »statistically significant« correlation does not guarantee analytically acceptable agreement between two methods of assay.
