Serum Levels of Carcinoembryonic Antigen and a Tumor-extracted Carcinoembryonic Antigen-related Antigen in Cancer Patients1

Levels of Carcinoembryonic antigen (CEA) and a tumor-ex tracted CEA-related antigen (TEX) were determined in sera from patients with carcinomas of the breast, colon, lung, head and neck, and a number of miscellaneous categories. The purpose of this study was to compare the levels of each antigen at various stages of malignant disease. Competitive radioimmunoassays developed in our laboratory were shown to be sufficiently specific to detect either of the.se two antigens independent of each other. The results indicate that: (a) with our specific assays, CEA was not significantly elevated in smoker controls, but TEX was elevated in 29% of smoker controls; (fa)TEX was equivalent to CEA as a tumor marker for colonie cancer. TEX was better than CEA as a marker for lung cancer and, based on limited data, there is a possibility that TEX is a significantly better tumor marker than is CEA in early lung cancer; (c) TEX was superior to CEA as a tumor marker for breast and head and neck cancers; (d) there is a strong indication that serial determinations of TEX can be used as effectively as CEA in the monitoring of disease progress. These preliminary results must be confirmed by increasing the number of cancer patients and including nonmalignant disease con trols. INTRODUCTION CEA3 was first described by Gold and Freedman (6, 7) in 1965 as a tumor marker for colonie cancer. Subsequently, a number of CEA-related antigens have been described (for review, see Refs. 8 and 14) including NCA (17), found in lung and spleen tissue, and TEX (10), coisolated with CEA from liver métastasesof colonie carcinoma. Structural studies on CEA, NCA, and TEX (5, 13, 14) indicate that NCA and TEX have identical NH2-terminal amino acid sequences, similar molecular weights (M, ~ 100,000), and nearly identical amino acid and carbohydrate compositions. Both differ from CEA in that CEA has a higher molecular weight (M, ~ 180,000), a higher per centage of carbohydrate (50 to 60% versus 30 to 35% for NCA and TEX), valine at position 21 in its NH2-terminal amino acid sequence in place of alanine for NCA and TEX, and no detect able methionine (NCA and TEX contain approximately 4 to 6 methionines). The development of a radioimmunoassay for 1Supported by Grants CA 16434 from the National Cancer Institute and CA 19163 from the National Large Bowel Program. The authors are members of the City of Hope Cancer Research Center. 2 To whom requests for reprints should be addressed. 3 The abbreviations used are: CEA, Carcinoembryonic antigen; NCA. nonspe cific cross-reacting antigen; TEX, tumor extracted CEA-related antigen; CMF, cyclophosphamide-methotrexate-5-fluorouracil. Received December 22. 1980; accepted March 11,1982. NCA by Burtin, von Kleist, et al. (16, 18) enabled them to evaluate the potential use of NCA as a tumor marker. They concluded that NCA was a very poor marker for cancer. Studies in our laboratory indicated that TEX possessed antigenic determinants found in CEA but not in NCA, thus sug gesting the possibility that TEX may behave differently from NCA as a tumor marker. In order to evaluate whether TEX was a better marker than CEA for various cancers, a pilot study was initiated at the City of Hope National Medical Center. In 331 cancer patients with a wide variety of carcinomas, the serum levels of CEA and TEX were determined by competitive radioimmunoassays in which a high degree of antigen specificity was exhibited. The results indicate that TEX may be a superior tumor marker to CEA in all types of cancer studied except for colonie cancer, in which case it was equivalent to CEA. It should be stressed that these results require confirmation by inclusion of larger numbers of patients with cancer and controls with nonmalignant diseases. MATERIALS AND METHODS Serum Samples. Serum samples from 100 healthy volunteers (59 nonsmokers, 41 smokers) were used as normal controls. Serial serum samples were drawn from 331 cancer patients who were seen at the City of Hope Medical Center over a 1-year period. Patient selection criteria consisted of histologically proven diagnosis of cancer. Samples were drawn serially at 2to 6-week intervals when possible and were stored at —20°. Voluntary informed consents were obtained prior to blood drawing. Radioimmunoassays. CEA and TEX used as assay standards were isolated by the perchloric acid method as previously described (1,10). Radioimmunoassays were performed by a triple-isotope, double-anti body method (2) utilizing 57Co as a supernatant marker (3). The CEA assay used goat anti-CEA as the primary antibody, and the TEX assay used rabbit anti-TEX. Under the assay conditions established, TEX did not interfere (was not detected as CEA) in the CEA assay up to 300 ng/ml, and CEA did not interfere in the TEX assay up to 600 ng/ml. Examples of the levels of interference are the following: TEX (1.5 ¡ig/ ml) is equivalent to CEA (8 ng/ml) in the CEA assay, and CEA (12 /¿g/ ml) is equivalent to TEX (50 ng/ml) in the TEX assay. In general, these assay conditions were suitable for independent measurements of CEA and TEX, since even for high values of one antigen in its assay the corresponding levels in the other assay were very low and could be corrected for if desired. Statistical Analysis. Serum levels of CEA and TEX are presented as mean ±S.E. Tests for equality of serum levels were made by 2-sided f tests using a pooled estimate of the variance. McNemar's test for matched samples, corrected for continuity, was used to test for equality of proportions positive for CEA and TEX.