An enzyme immunoassay procedure for cancer antigen 125 evaluated.

The performance of a new enzyme immunoassay (EIA) procedure (Abbott Labs.) for cancer antigen 125 (CA 125) met or exceeded the manufacturer's claims for all analytical variables examined. Overall correlation with results obtained with a radioimmunoassay (RIA) were good. However, near the decision thresholds typically chosen to define a positive result for ovarian carcinoma, EIA results were 10 to 20 arbitrary units/mL less than the RIA results. At specific decision thresholds, therefore, the sensitivities and specificities of the EIA and RIA procedure differed. Adjusting the decision thresholds gave a similar optimum efficiency for each procedure: EIA, 82.9% (decision threshold, 35 units/mL); RIA, 83.4% (decision threshold, 54 units/mL). Receiver-operating characteristic curves showed that the two procedures' ability to distinguish patients with active ovarian carcinoma from those with disease in remission was the same.

Evaluation of a solid-phase enzyme immunoassay for human choriogonadotropin beta subunit.

We evaluated a quantitative solid-phase enzyme immunoassay for human choriogonadotropin beta subunit (beta-HCG) with anti-beta-HCG:horseradish peroxidase conjugate, recently marketed by Abbott Laboratories. We compared results on 56 patients' serum specimens, obtained mostly for followup of neoplastic disease, with those by a competitive radioimmunoassay kit. The correlation was good, the differences being of little clinical significance. Linear regression in the low and intermediate ranges gave a slope of 0.93, a y-intercept of 0.34, and a correlation coefficient of 0.97. Precision studies yielded an interassay CV of 6.4% in the intermediate range and 13% in the low range. Sensitivity was 0.69 int. unit/L. Cross reactivity was 1 to 2% with specimens fortified with lutropin or follitropin. The only substantial problem was with linearity in the upper part of the standard curve, especially in the interval, 100-200 int. units/L. This problem is obviated by adequate sample dilution.