Comparison of five immunoassay procedures for the ovarian carcinoma-associated antigenic determinant CA 125 in serum

Interassay variability in CA 125 values was studied in 77 serum samples (covering a range of CA 125 values between 8.9 and 310 arb. U/ml as measured by the original Centocor RIA) using three 125I-labeled RIA kits (Centocor, Byk and Cis) and two enzyme-labeled immunoassays (Abbott and Roche). Taking the Centocor RIA as a reference, orthogonal regression equations resulted in slopes varying between 0.74 and 1.35, with y-axis intercepts varying between -6.5 and +6.2, and correlation coefficients ranging from 0.88 to 0.94. Compared with the results of the Centocor RIA, the EIAs of Abbott and Roche gave overall lower CA 125 values, whereas the Cis and the Byk RIAs gave higher assay results. At the 35 arb. U/ml Centocor cut-off, serum levels with the other assays varied between 23 and 53 arb. U/ml. The 65 arb. U/ml cut-off level corresponded with CA 125 serum levels between 45 and 94 arb. U/ml. When classifying CA 125 values in three clinically relevant categories based on Centocor RIA results, ('normal' < or = 35 arb. U/ml, 'slightly elevated' > 35- < or = 65 arb. U/ml and 'elevated' > 65 arb. U/ml), discordances ranged from 26% with the Cis RIA to 40% utilizing the Byk RIA. The five CA 125 assays tested do not give equal assay results. As a consequence, the interpretation of CA 125 serum concentrations should be done with caution in disease monitoring and in the assessment of ovarian masses, especially when using different serum assays.