Abstract
Carcinoembryonic antigen (CEA) was detected in 180 preoperative malignant serum samples, or from the first benign serum sample if more than one sample was taken by the Farr assay using reagents provided by the Montreal laboratory of Gold and Freedman and in 148 plasma and 178 serum samples by a solid‐phase assay using reagents provided by the Cancer Diagnostic Laboratory of Hoffmann‐La Roche, Inc. Disposable, polystyrene tubes were coated with anti‐CEAγG in the solid‐phase assay. These were then used along with ^125^I‐CEA to construct an antibody dilution curve, standard inhibition curve and to test plasma or serum samples directly for CEA. In general, levels of CEA were higher for all diagnostic categories by the Farr assay than by the solid‐phase assay. For CEA levels greater than 10 ng/ml, samples from colorectal cancer patients were clearly separable from other samples by both techniques. Thus, levels greater than 10 ng/ml were found by the Farr and solid‐phase assays done on serum and by the solid‐phase assay done on plasma respectively in 39%, 23% and 13% of patients with colorectal cancer, and in 14%, 4.5% and 1% of patients with other diseases. Aliquots from 123 blood samples were tested by both the Farr and the solid‐phase assays. Serum was studied in all 123 by the Farr assay. Fifty‐seven of the solid‐phase assays were done on serum and the remaining 66 were done on plasma. The results showed good agreement (77% overall) for the two assays when serum results were considered, but lesser agreement (59% overall) when serum results by the Farr assay were compared with plasma results by solid‐phase. The serum/plasma disagreement noted between the Farr and solid‐phase assays was largely found in samples from patients with colorectal cancer (45% agreement), and appeared to be due mainly to samples from patients with Duke's A or B colonic cancer. Antibody dilution and standard inhibition curves done using reagents from the two laboratories in both the Farr and solid‐phase assays indicated that the CEA from Montreal detects antibody sites on both anti‐CEA preparations which are not available to the CEA from Roche. Further, the unlabelled CEA from Roche appears to be more similar antigenically to the Montreal ^125^I‐CEA than unlabelled Montreal CEA itself. This suggests that iodination of CEA may have altered the anti‐genie character of the molecule. It is suggested that this alteration, as well as the use of CEA preparations of higher specific activity than those used initially, might account, at least in part, for the apparent loss of specificity of recent “CEA tests” for colorectal cancer as compared to the earlier report from Thomson and colleagues.