Determination of carcinoembryonic antigen (CEA): comparison of two different assays

A comparison was made of CEA determination by the CEA-RIAKIT and our slightly modified CEA-Roche assays. The normal levels (mean + 2 SD) found in 63 blood donors were 2.3 microgram/1 and 3.3 microgram/1, respectively. The inter-assay reproducibility was similar in the range below 15 microgram/1, with variations from 0.3 microgram/1 (1 SD, less than 5 microgram/1) to 1.4 microgram/1 (1 SD, 10--15 microgram/1). Both methods measured CEA concentrations down to 1 microgram/1. The ability of the two assays to discriminate between positive and negative values in patients with urogenital and gastrointestinal cancer was similar. The response to recurrences of colorectal carcinoma was also similar, and the CEA fluctuation was parallel after surgery. The CEA-Roche assay generally showed good correlation between measured and theoretical values at both low and high levels. Under our test conditions, the disparity between indirect and direct values was 14 microgram/1 +/- 8 (n = 4) in the range 18--32 microgram/1, and this must be taken into consideration in interpreting the CEA-Roche test. The CEA-RIAKIT measured too low values as compared to both the CEA-Roche and the theoretical values. The discrepancy increased with increasing CEA values, and the ability to distinguish between high CEA concentrations (greater than 15 microgram/1) was poor. The explanation seems to be that the anti-CEA sera used in this assay show a lower affinity for plasma CEA than for tumour-extracted CEA. Immunochemical differences between sample CEA and CEA used for calibration of the standard curves must be considered in evaluation of CEA assays.