Comparison of RIA and IRMA methods for measurement of carcinoembryonic antigen (CEA)

We present data comparing the Roche CEA radioimmunoassay (RIA) utilizing ultrafiltration and the Abbott CEA immunoradiometric assay (IRMA) methods for standard curve sensitivity, analytical reproducibility, recovery (pre and post extraction), dilution linearity, and patient correlation. The Roche intra- and inter-assay precision figures for this assay were 8% and 15%, respectively, for a CEA concentration of 4.0 ng/mL. The Abbott assay gave comparable precision values of 9% and 16%, respectively, for a CEA concentration of 2.0 ng/mL. Recovery for plasma spiked with CEA stock preparations was commercial-source-dependent when assayed by the Roche assay, but independent when measured by the Abbott assay. The Roche assay did not recover both preparations quantitatively, while the Abbott assay did. Plasma dilution studies over a wide range of CEA concentrations gave linear results for both assays provided the Roche assay utilizes the extraction method with dialyzate dilution for CEA values greater than 20 ng/mL. A difference (60%) was observed in linear regression slopes of the measured values between the Roche "indirect" and "direct" assays. Exposure of CEA to perchloric acid seems to play a critical role in recovery of CEA material when assayed by the Roche method. A large number of the patient samples assayed gave widely discrepant results when comparing the RIA to the IRMA methods. The potential significance of these discrepancies is discussed.