TIBS
- June 1979
121
Reviews Carcinoembryonic antigen and related glycoproteins ai tumor markers C. M. Sturgeon Human tumors often secrete abnormal substances whose measurement in body fluids may yield valuable information about the cancer’s growth. The most important of these secreted molecules is the carcinoembryonic antigen, the detection and measurement of which has great potential value in the management of cancer patients.
The realization that early detection and treatment of malignant disease can greatly improve prognosis in cancer patients has inspired considerable research, and much effort has been directed towards the biochemical investigation of substances secreted by cancer cells. Measurement of these compounds may yield information of clinical value in diagnosing, localizing or monitoring tumor growth. ‘Tumor markers’ are stable molecules secreted by cells of a specific morphological type. Present in serum or urine, their concentration is, ideally, proportional to tumor burden (the number of viable tumor cells in the patient) and shows a prompt response to changes in tumor size. Tumor markers which have been described include fetal antigens, hormones, enzymes, glycoproteins and various metabolic products. Demonstration of their existence, however, does not ensure their applicability: a sensitive assay method must be available, the presence in body fluids of similar substances produced by normal tissues must be investigated to determine whether cross-reactions occur in the assay method used, and the effects of possible autoantibody production directed at the tumor marker must be studied. No tumor marker fulfils all these requirements as yet, although for trophoblastic tumors, human chorionic gonadotrophin is close to ideal, as is a-fetoprotein for hepatomas and teratomas [I]. A marker with much wider applicability and versatility, however, is the carcinoembryonic antigen (CEA), C. M. Sturgeon is a Research Associate at the Department of Clinical Chemistry, University of Edinburgh, U.K.
which is frequently found in high concentration in the sera of patients suffering from colorectal and other cancers. CEA: biochemical properties CEA was first reported to be present in extracts of tumor tissue and of normal fetal colon [2]. It was subsequently characterized as a glycoprotein, molecular weight approximately 200,000, containing 40-507; carbohydrate. It has a single polypeptide chain, and the sugar residues are thought to extend in about 90 units, 10 residues long, from the protein backbone [3]. Up to 90% of the carbohydrate can be removed chemically using a sequence of periodate oxidation, borohydride reduction and mild acid hydrolysis [4]. The deglycosylated product retains some immunoreactivity, suggesting that the immunological determinant resides in the protein portion of the molecule. CEA is usually prepared from liver metastases to primary colorectal carcinomas which often contain particularly high levels of the antigen. Single metastases may contain up to 1 g of CEA [5]. The content varies markedly from tumor to tumor, however, and may even vary within a single tumor. Purification is usually accomplished by solubilization in perchloric acid. After exhaustive dialysis of the extract followed by gel chromatography, ion-exchange and lectin chromatography steps are sometimes included in the purification procedure. Use of these techniques readily demonstrates the heterogeneity of ‘CEA’, as numerous molecular variants are separable. These variants differ chiefly with respect to carbohydrate content and charge, and their composi-
tions in an> CEA preparation will depend on tumor source and method of preparation. It is probably realistic to consider CEA as a family of closely related glycoproteins rather than as a single species. Immunological activity is usually monitored during purification using radioimmunoassay with reagents shown to react similarly to those originally developed
[1,61. CEA : measurement CEA is satisfactorily measured by radioimmunoassay, a technique which is very sensitive (detection limit 2.5 rig/ml), precise and well-suited to use in clinical laboratories as large numbers of samples can be conveniently handled. Establishing a radioimmunoassay for CEA requires particular care, however. The quality of any radioimmunoassay is determined by the purity of the labelled antigen, by the specificity of the antiserum, and by the efficiency of the method of separating ‘bound’ from ‘free’ antigen to complete the assay. For CEA the last presents no difficulties : there are a number of successful separation methods used for CEA radioimmunoassays, including use of double antibody, ammonium sulphate or zirconyl phosphate precipitation steps. Obtaining a highly purified antigen for labelling, however, is more difficult. The heterogeneity of CEA is such that selecting the most antigenic fraction from a lectin-affinity or ion-exchange column run may be difficult. It is advantageous to examine the behaviour of the fractions chosen for labelling on gel filtration after iodination in order to assesstheir suitability for use in the radioimmunoassay [7]. @I Elsevier/North-Holland
Biomedical
Press 1979
122 Antisera to CEA preparations are readily raised, usually in rabbits or goats. It is now generally accepted that the CEA molecule carries a number of immunological determinants, all of which are likely to be located on the protein moiety, although evidence for this has not yet been satisfactorily obtained. At least one determinant is common to a smaller molecular weight (approx. 60,000) molecule (NCA = ‘non-specific cross-reacting antigen’) present in normal tissue, and to a second similarly sized antigen (NCA-2) which is extracted from normal faeces and meconium [8]. Another determinant is shared by CEA and NCA-2, but not by NCA. Each of the three antigens (CEA, NCA and NCA-2) possesses its own antigenic determinant as well. The unique CEA determinant is the one postulated to have ‘cancer-specificity’, and it is the antibodies to that particular determinant which are required in an antisera suitable for CEA radioimmunoassay. However pure the CEA used as immunogen, some of the antibodies raised will inevitably recognize the determinants of the non-specific cross-reacting antigens. To remove as many of these undesirable antibodies as possible, anti-CEA sera are usually adsorbed with perchloric acid extracts of normal serum, spleen, lung and colon. Additional difficulties arise because normal tissues contain very small amounts of CEA itself. Consequently, judicious choice of ‘when to stop’ adsorbing an antiserum is necessary, as it is possible to remove almost all antibodies from the antiserum. No two antisera have identical populations of antibodies, so it can be readily appreciated that the necessity of adsorbing the antiserum further compounds the difficulty of obtaining a standard antiserum for radioimmunoassay of CEA. As a result, reagents for the assay may differ markedly from laboratory to laboratory, making correlation of interlaboratory serum CEA values difficult. The situation has been somewhat improved in the United Kingdom by the adoption of a national standard for CEA, to which numerical values may be referred. CEA and the clinician The early hope that CEA measurement would provide a powerful diagnostic tool for colorectal cancer aroused high expectations which have not been fulfilled. Some of the diseases in which serum CEA levels are elevated above ‘normal’ (i.e. 2.5 rig/ml) are listed in Table I [9]. The percentage of pancreatic carcinomas in which CEA is elevated (92 “/,) is even higher than that for
TIBS
-
June 1979
colorectal cancer (73 %). Benign disorders Serum CEA of the gut, including diverticular disease 120-concentration tnglml) and ulcerative colitis, may also result in loo’ Abdominoelevated CEA levels. It is interesting to perineal resection / note that in patients suffering from these 80 diseases. high CEA levels usually coincide X-ray therapy with periods when clinical symptoms are 60 ; for one week f ,/l severe, which may be‘ significant as these 40 h, diseases are suggested to be precancerous / ,,,,*A/ conditions. The lack of diagnostic speci20 ‘i’ ficity of CEA is readily apparent: in certain cases CEA measurement may help to 1~ 1 2 3 -k-g 6 7 8 confirm a diagnosis, but its value is very Time (months) limited. Evaluation of CEA measurement Fig. 1. CEA profile for patient L. admitted to as a screening procedure in non-cancer hospital with knohv cancer of the rectum. CEA clinics has recently been reported [JO]. The levels measured after surgery decreuse during the increased indicating authors suggest that factors such as first month, then gradually incomplete removal of tumor. X-ray therapy failed improving the specificity of the CEA assay, to halt the spread of the tumor. selecting high-risk populations for screening, organizing a carefully designed clinical follow-up plan and assessing the costoperation site. Measurement of CEA can benefit ratio should be considered before similarly be used to monitor the progress of employing CEA as a screening tool for patients during and following treatment by cancer. It is chiefly as a prognostic aid and X-ray therapy or chemotherapy. Successful when used in post-treatment surveillance treatment is followed by a decline in CEA of cancer patients that CEA measurement levels. Continued elevation of the level may be of considerable importance to the suggests that the tumor is not responding clinician. to the regime employed and can prompt Long-term follow-up results from nu- alteration of the treatment. merous laboratories have shown that after In those patients who have elevated successful surgery a raised pre-operative serum CEA concentrations, levels have CEA level will fall to within normal levels. been shown to reflect tumor size. Grossly If there is residual tumor the CEA level elevated levels (> 100 rig/ml) are usually often remains elevated, a warning to the indicative of metastatic disease or very surgeon which may be sufficient indication serious liver damage. Treatment in these for a ‘second-look’ operation to determine cases often fails to decrease tumor burden whether tumor tissue remains at the site of or CEA levels significantly, although it the operation. Persistently raised CEA may prevent further spread of the disease. levels after surgery may also indicate a CEA may aid the clinician considerably second tumor mass remote from the by yielding information about the efficacy of the treatment used. It is also of use in TABLE I the long-term follow-up of patients after Incidence of raised plasma CEA levelsin various successful therapy. In patients whose CEA disorders levels have fallen to within normal levels Incidence (70) after treatment, a gradual increase in these of raised levels with time strongly suggests recurplasma CEA rence. Serial CEA measurements are of Disease levels much greater value than single determinaCarcinoma of tions, as can be appreciated by considering Colon and rectum 73 CEA ‘profiles’ for two patients. Patient ‘L’ 92 Pancreas was admitted with known cancer of the Liver 67 rectum. The serum CEA value was high Bronchus 72 and did not fall to normal levels after Breast 52 Uterus 53 operation (Fig. I), but gradually increased Ovary 36 with time. X-ray therapy proved ineffective, Ulcerative colitis and although there was a small initial drop in Crohn’s disease 21 CEA level, and the patient died two months Cirrhosis and alcoholic liver after the last serum CEA level was 42 disease Chronic bronchitis and measured. Patient ‘MC’ was admitted to emphysema 25 hospital in January 1976. After suffering 7 Fibroadenosis from ulcerative colitis for twenty years he had developed a malignant stricture in the From data compiled by Munro-Neville and colon. CEA measurements were normal Cooper [9].
TIBS - June IF79
123
and after apparently successfui resection of the colon (Fig. 2). The patient returned to work and remained heaithy for more than a year. Twelve months after his operation, however, his CEA levels had risen to 8 rig/ml. His serum was retested three months later and the CEA level was found to be 64 ngiml, and a month and a half later it had farther increased to 70 ngjml. At this time a second-look operation was performed on the patient, who was complaining of some abdominal clinically he remained pain, although asymptomatic. The operation revealed gross recurrence on the walls of the bowel. The tumor was inoperabls, but the cancerous sites were ‘mapped’ at operation so allowing the patient to receive radiotherapy. during
CEA : future developments The fact that rising CEA titres may precede the onset of clinical symptoms by many months is not fully exploited at present since the medical technology for dealing with barely detectable tumor masses is not yet available. Early detection and treatment of cancer is highly desirable, but until the clinician can palpate the tumor or can accurately locate it, there is little that can be done for the patient. It is likely that until more sophisticated surgical, radiative or chemotherapeutic treatment of early cancerous growth is possible the full potential of CEA as a monitor of tumor growth will not be realized. It is possible that CEA or other tumor markers may in fact aid in the early location of tumor tissue in the future. A Second-look reveals gross
operation recurrence
Serum CEA , concentration 7os (rig/ml)
I f d’
Still clinically asymptomatic; patient complaining of some discomfort
60
& w’
50 40’ 3r Resection 20, malignant stricture 10
of colonic
.
7 i-
*+*-a-
1
j
2l’.
.
~,
7 Time
Fig. 2. hospital colon. months
12 (months)
~-~-.~-
16
17
CEA profile for patient MC‘, admitted to for excision of a malignant stricture in the CEA levels remained normal until twelve after operation, and then increased rapidly.
recent report describes the successful use of radiolabelled antibodies to CEA for the detection and localization of diverse cancers by external photoscanning [I I]. A radiolabelled and affinity-purified goat IgG having 70 “/< immunoreactivity against CEA was administered to 18 patients with a history of cancer. Each patient received approximately 1 mCi of 131K-labelled IgG. As expected, blood-pool background radioactivity was considerable, owing to circulating CEA, and ordinary photoscans were therefore difficult to interpret. By using a computer, subtraction of radioactive blood-pool agents from the antibody’s 13r1 activity could be effected. In almost all cases studied, tumor location could be demonstrated at 48 hours after injection. Scans vvere negative for patients without demonstrable tumors or with tumors apparently devoid of CEA. Circulating antigen levels of up to 350 rig/ml did not prevent successful tumor imaging after injection of the radioantibody. Four cases gave photoscan evidence of tumors that were not substantiated by clinical findings until operation or autopsy was performed. This work, still in its infancy, has exciting implications for the future. It demonstrates that successful radioimmunolocalization of tumor does not require a truly tumor-specific antigen and suggests that use of CEA radioantibodies may be of future importance, especially if administration of smaller quantities of radioactivity becomes feasible. Despite considerable research effort, no wholly satisfactory explanation for the reappearance of CEA or other fetal antigens in the cancerous state has yet been proposed. The biological function of these molecules in fetal or cancerous tissue has not been explained either. It appears to be common, perhaps as a result of genetic derepression, for embryonic substances, placental or fetal in type, to be produced in malignant tissue. A similar argument may explain the presence of small amounts of CEA in normal tissue: because repression of a gene is never wholly complete in practice, one would expect to find traces of embryonic antigens in normal tissue. It has been suggested that CEA and related antigens may contribute to the protection of the mucosae against external aggressions. The high carbohydrate content of CEA, its heterogeneity (sometimes attributed to varying levels of sialic acid in the molecules), and the fact that some CEA preparations appear to share reactivity with blood-group substances, has prompted the proposal that CEA may be a deviant form of a blood-group substance
[12]. The lack of similarity of the amino acid compositions of the blood-group substances and CEA tends to refute this, but it is possible that CEA production is connected with the inappropriate action of glycosyl transferases on a normal serum glycoprotein. The similarities between the amino-acid compositions of CEA and NCA, and the distinct differences in their degrees of glycosylation have been noted, and extensive homology has been demonstrated between the NH,-terminal sequence of CEA and NCA [l3]. Of the 26 amino acid residues whose sequence has been determined in NCA, only one differed from those in CEA. It has therefore been suggested that the gene coding for CEA or NCA has undergone a very recent duplication. This perhaps clarifies the relationship between NCA and CEA, and may help to explain their immunological cross-reactivity. When satisfactory explanations can be provided for the reappearance of CEA in cancer, a much greater understanding of the nature of the disease state will have been achieved. Although there is still much to be learned about the physiological role of CEA, and many questions yet to be answered about its biochemical character, CEA is already without doubt the most useful and interesting of the tumor marker substances. References Bagshawe, K. D. and Searle, F. (I 977) Essays in Medical 2
Gold, Exp.
3
Biochemistry
Med.
Slayter,
5 6
H. S. and Coligan,
Bessell, E. M., Thomas, P. and Westwood, J. H. (1975) Carbohydr. Res. 45, 257-268 Pritchard, D. G., Todd, C. W. and Egan, M. L. (1978) Methods Cancer Res. 14, 55-85 Thomson, D. M. P., Krupey, J., Freedman, S. 0. and Gold, P. (1969) Proc. Nat. Acad. Sri.
U.S.A.
64, C.
Edinburgh
9
161-167 M.
J.
(1978)
Burtin,
23, 3 19-325. P. (1978) Ann.
Pasteur)
129C.
185-I
Munro-Neville, Ann.
Clin.
IO Chu,
State II
J. E. (1975)
14, 2323-2330
7 Sturgeon, 8
S. 0. (1965) J.
12 I, 439.-462
Biochemistry 4
3, 25-73
P. and Freedman,
Immunol.
Surg. (Inst.
98
13, 283-305
and
J. Med.,
Murphy,
G.
P. (1978)
N.Y.
879-882
Goldenberg, Bennett,
Coil.
A. and Cooper, E. H. (1976)
Biochem.
T. M.
R.
D. M., DeLand,
S., Primus,
F., Kim,
F. J., van
Nagell,
E.,
J. R.,
Estes, N., De Simone, P. and Rayburn, (1978)
Engl.
N.
12 Simmons,
Cancer 13 Engvall,
(1978) 1674
D. A.
Res. E.,
Proc.
J. Med. R.
298,
and
P.
1384-1388
Perlmann,
P. (1973)
33, 313-322 Shively,
Nut.
Acad.
J. E.
Sri.
and
U.S.A.
Wrann,
M.
75, 1670-