Biochimica et Biophysica .4cta, 417 (1974) 123-152 © Elsevier Scientific Publishing C o m p a n y , A m s t e r d a m - Printed in The Netherlands BBA 87016

CARCINOEMBRYON1C

MOLECULAR

BIOLOGY

ANTIGEN

AND

*ABRAHAM FUKS, C H A I M BANJO, and ***PHIL G O L D

(CEA):

CLINICAL

SIGNIFICANCE

* * J O S E P H S H U S T E R , S A M U E L O. F R E E D M A N

Division of Clinical Immunology and ,4llergy, Montreal General Hospital, and the McGill University Medical Clinic of the Montreal General Hospital Research Institute, Montreal (Canada) (Received N o v e m b e r 15th, 1974)

CONTENTS I.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

124

11.

Animal t u m o r antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. T u m o r antigens in experimentally induced animal t u m o r s . . . . . . . . . . . . . B. The induction of experimental animal t u m o r s . . . . . . . . . . . . . . . . . .

124 124 125

III. H u m a n t u m o r antigens . . . . . . . . . . . . . A. T u m o r immunity as indicator of tumor-specific B. The demonstration of h u m a n t u m o r antigens animals . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . antigenicity . . . . . . by the immunization . . . . . . . . . . .

. . . . . . . . . . . of xenogeneic . . . . .

IV. The carcinoembryonic antigen (CEA) of the h u m a n digestive system . . . . . . . . . A. Techniques for the demonstration of C E A . . . . . . . . . . . . . . . . . . . B. Distribution and t u m o r specificity of CEA . . . . . . . . . . . . . . . . . . . C. Cellular localization of the C E A . . . . . . . . . . . . . . . . . . . . . . . . D. Cellular and h u m o r a l i m m u n e responses to CEA . . . . . . . . . . . . . . . . E. Circulating C E A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Metabolism of C E A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. R a d i o i m m u n o a s s a y s for C E A . . . . . . . . . . . . . . . . . . . . . . . . . V. Purification and physicochemical properties of CEA . . . . . . . . . . . . A. Purification of CEA . . . . . . . . . . . . . . . . . . . . . . . . B. Physical properties of C E A preparations . . . . . . . . . . . . . . . 1. Molecular size . . . . . . . . . . . . . . . . . . . . . . . . . 2. Molecular charge . . . . . . . . . . . . . . . . . . . . . . . . 3. Miscellaneous physical parameters . . . . . . . . . . . . . . . . C. Structure of CEA . . . . . . . . . . . . . . . . . . . . . . . . . 1. Carbohydrate composition . . . . . . . . . . . . . . . . . . . . 2. Structure of the carbohydrate moiety . . . . . . . . . . . . . . . a. Acid hydrolysis of CEA . . . . . . . . . . . . . . . . . . . . b. Enzymatic hydrolysis of CEA . . . . . . . . . . . . . . . . . c. Periodate oxidation of C E A . . . . . . . . . . . . . . . . . . d. Carbohydrate c o m p o n e n t s of glycopeptides . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

* Fellow of the Medical Research Council of Canada ** Clinical Research Associate of the National Cancer Institute of C a n a d a *** Associate of the Medical Research Council of Canada

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

126 126 126 127 127 128 128 129 130 131 131 133 133 134 134 134 135 135 135 136 136 137 137 137

124

D. E.

F. G.

3. Amino acid composition of CEA . . . . . . . . . . . . . . . . . . . . 4. Primary structure of the protein moiety of CEA . . . . . . . . . . . . . Nature of the carbohydrate-protein linkage in CEA . . . . . . . . . . . . . Immunochemical studies of CEA . . . . . . . . . . . . . . . . . . . . . 1. Binding studies between 1251--CEA and various antisera . . . . . . . . . . . 2. Specific lectins . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Comparative inhibition studies in the ~zsI--CEA anti-CEA system . . . . . . . 4. Materials cross-reacting with CEA . . . . . . . . . . . . . . . . . . . The tumor-specific grouping of the CEA molecule . . . . . . . . . . . . . . Nature of the circulating substance(s) measured by radioimmunoassay . . . . . . .

. . . .

. . . .

. . . . . .

138 139 139 139 139 140 142 142 143 145

VI. Clinical applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Diagnosis of cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Use of assay for the surveillance of cancer patients . . . . . . . . . . . . . . .

145 147 148

VII. Future perspectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

149

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

149

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

149

I. INTRODUCTION The review that follows will deal primarily with the i n f o r m a t i o n presently available c o n c e r n i n g the c a r c i n o e m b r y o n i c antigen (CEA) of the h u m a n digestive system. I n order to keep this subject in perspective, however, the concept of t u m o r antigens in general will be briefly surveyed.

1I. ANIMAL TUMOR ANTIGENS

IIA. Tumor antigens in experimentally induced animal tumors Investigations based on the rejection of experimentally induced tumors, transplanted between m e m b e r s of highly i n b r e d or syngeneic rodent strains u n d e r appropriate circumstances, have established the existence of tumor-specific antigens in m a n y animal tumors. The exchange of n o r m a l tissue grafts between syngeneic r o d e n t s does not lead to the rejection of the t r a n s p l a n t e d tissue because b o t h the host a n d recipient express identical t r a n s p l a n t a t i o n antigens. Hence, rejection of a t u m o r graft between such animals can be evoked only by new, tumor-specific transplanta t i o n antigen(s) which have arisen in the t u m o r tissue during the process of m a l i g n a n t t r a n s f o r m a t i o n [1-3]. It should be noted that the term "tumor-specific t r a n s p l a n t a t i o n antigen" is a semantic convenience. Hence, although the tumor-specific t r a n s p l a n t a t i o n antigen can stimulate a protective i m m u n e response on the part of the host, it is unlikely that such i m m u n o g e n i c i t y is the p r i m a r y f u n c t i o n of these molecules. There is a far Reprint requests should be addressed to P. G., Rm. 7135, Montreal General Hospital, 1650 Cedar Ave., Montreal, Quebec, H3G 1A4

125 greater likelihood that tumor antigens play well-defined, but yet undetermined, roles in maintaining the integrity of cancer cells. The tumor-specific transplantation antigens are constituents of tumor cell surface membranes, and are capable of stimulating both humoral and cell-mediated immune responses on the part of the tumor-bearing host [4]. The interaction of the two forms of immune responsiveness against tumor antigens in both animals and humans remains incompletely defined. A number of factors should, however, be considered. First, a tumor can continue to grow in the face of specific antitumor immune responses. From studies performed in vitro, it would appear that the cellmediated immune reaction is cytotoxic to cancer cells [5,6]. However. this cellmediated cytotoxicity may be inhibited, at the lymphocyte level, by circulating tumor antigen-antitumor antibody complexes or by free antigen alone, or blocked at the target cell level by free antibody [7]. Furthermore, uncomplexed, humoral antitumor antibody may, under certain circumstances, serve as unblocking factor, reversing the action of certain blocking materials [8]. The role of these, and various other factors, and their interaction in vivo, has still to be elucidated, liB. The induction of experimental animal tumors Rodent tumors may be induced by a wide variety of chemical carcinogenic agents (e.g., polycyclic hydrocarbons, nitrosamines), or oncogeneic viruses of either the D N A (e.g., polyoma, SV40, adenovirus) or RNA (avian and murine leukemiasarcoma viruses, mammary tumor viruses) classes [2,3]. A good deal of evidence indicates that these mutagens are also able to modify the immunologic status of the host so as to suppress the immune response which might otherwise inactivate the inducing agent or cause a rejection of transformed cells bearing tumor-specific transplantation antigens. Thus, many of the chemical carcinogens and oncogenic DNA viruses may produce a non-specific immunologic depression, while the vertically transmitted RNA viruses may produce a full and specific immunologic tolerance on the part of the host, Until recently, it was believed that each tumor induced by a chemical carcinogen contained a unique tumor-specific transplantation antigen that was distinct from any other tumor induced even by the same agent [9]. However, during the last few years, it has been shown that, in addition to their unique tumor-specific transplantation antigens, chemically-induced tumors also express tumor-associated fetal antigens which are common components of different chemically-induced tumors [7]. For reasons that are not well understood, the fetal antigens associated with these tumors apparently do not function effectively in stimulating a tumor graft rejection reaction. Tumors produced by the same DNA virus manifest common transplantation antigens [2]. In addition, other neoantigens arise which do not evoke a rejection reaction [7]. Such tumors usually lack virion antigens and do not release infective virus particles. On the other hand, RNA virus-induced tumors contain both transplantation antigens and virion antigens common to cells transformed by the same virus [2]. Such cells usually release infective virus, except under circumstances in

126 which the virus is defective. Certain of the RNA virus-induced and DNA virusinduced tumors have also been found to express fetal antigens, some of which are capable of evoking a tumor rejection response [7].

IlI. HUMAN TUMOR ANTIGENS The lack of a syngeneic donor-host relationship, and the moral and ethical restraints associated with tumor transplantation in humans, have virtually precluded this type of approach to the study of human cancers. A variety of alternative experimental designs has, therefore, been utilized in this aspect of human cancer immunology. IliA. Tumor immunity as an indicator of tumor-specific antigenicity The demonstration that a tumor-bearing patient is able to mount an immune response against his tumor is highly suggestive of tumor-specific antigenicity. Such immunity may be manifested by the development of either humoral antitumor antibodies, or by the occurrence of a demonstrable cell-mediated immune response. Quite frequently, both forms of immune reactivity are observed when they are sought by appropriate means [4]. Applying the criterion of immune responsiveness as an index of tumor-specific antigenicity, virtually every human neoplasm which has been adequately studied appears to contain such tumor-specific antigens [4]. Perhaps most thoroughly investigated from this point of view have been Burkitt's tumor and a variety of other lymphomas, various types of sarcomas, carcinomas of the colon, nasopharynx, urinary bladder, breast, lung, kidney, testis, endometrium, ovary, and malignant melanomas [4,10]. IIIB. The demonstration of human tumor antigens by the immunization of xenogeneic animals Although the phenomenon of tumor immunity as an indicator of tumor antigenicity provides some insight concerning host-tumor interaction, it does not readily allow for the definition of the nature of the tumor antigens involved. On the other hand, while studies dealing with the detection of human tumor antigens by the immunization of xenogeneic animals are fraught with the problems outlined below, they do provide a means for examining the nature of the tumor constituents which may be detected. It is in this manner that a number of materials normally produced in significantly detectable quantities only by normal fetal tissues have been found in human cancers. This process, alternatively termed "retrodifferentiation", "derepressive-dedifferentiation", or "antigenic reversion" appears to account for the synthesis of such materials as the carcinoembryonic antigen (CEA) of the human digestive system, alphal-fetoprotein and a number of other human tumor antigens. In addition, similar experimental approaches have allowed observations which indicate the

127 existence of tumor-specific antigens in human leukemic cells [11] and Hodgkin's tumor tissue [12].

IV. THE CARCINOEMBRYONIC ANTIGEN (CEA) OF THE HUMAN DIGESTIVE SYSTEM Two major difficulties arise in studies involving the immunization of xenogenic animals with human tumor tissue. First, all human tumor extracts contain large quantities of normal tissue +'contaminants" which occur on both the cellular and subcellular levels. The use of such material in heteroimmunization invariably evokes a predominantly antinormal tissue antibody response in the experimental animal host. Second, and more important, the validity of comparing the antigenic content of any tumor tissue under consideration with so-called normal control tissue, taken from noncancerous patients, is open to serious question. Under these circumstances, it is virtually impossible to distinguish tumor-specific antigens from normal, individual-specific, alloantigenic differences between the donors of normal and tumor materials.

IVA. Techniques for the demonstration of CEA In order to overcome these problems, adenocarcinoma of the human colon was chosen for study. From the experimental standpoint, this tumor has many advantages because of its limited ability to invade intramurally. The mucosa of surgically resected specimens taken more than 6 or 7 cm either proximal or distal to the site of the visible tumor in the gross, is free of tumor. Thus, both normal and tumor tissues may be obtained from the same patient, eliminating the problem of allogeneic differences between the tissue donors. Antitumor antisera were initially prepared in rabbits and rendered tumorspecific in one of two ways [13]. First, antiserum raised in adult rabbits against an extract of tumor tissue was absorbed with an excess of the corresponding normal tissue preparation to remove all of the antinormal tissue antibody activity from the antiserum in question. In a second series of investigations, neonatal rabbits were given a course of injections of normal tissue extract in an attempt to induce a state of immunologic tolerance to this material. These animals were subsequently immunized with corresponding tumor extract during normal life in order to induce a tumorspecific antibody response. The antisera prepared by both techniques were tested for their content of antibodies by a number of different immunochemical procedures, including precipitation in agar gel (Ouchterlony reactions and immunoelectrophoresis), bis-diazotized benzidine hemagglutination, passive cutaneous anaphylaxis in mice, and immunofluorescence. The results obtained by all of these serologic techniques were in accord. Moreover, the data resulting from experiments employing antiserum preparations obtained by the tolerance technique were, with minor variations, in agreement with those obtained using antiserum prepared by absorption.

128 Initial studies employing pools of colon cancers, and subsequent investigations in which many hundreds of individual specimens were examined, revealed the same tumor-specific component in every colonic tumor [14]. Numerous experiments excluded the possibility that the antibodies giving rise to the tumor-specific antigenantibody interaction were directed against either the bacterial flora of the bowel, or the usually high concentrations of fibrin frequently found in malignant tumors.

IVB. Distribution and tumor spec~city of CEA Utilizing the colon cancer system as a model, a study was undertaken to determine whether comparable components are present in other human tissues. Approximately 3000 tissue specimens from all parts of the human body, encompassing both normal and a variety of histopathological states, were examined by direct precipitation and precipitation inhibition in agar gel. It was found that the same constituent was detectable in all test specimens of primary and metastatic cancerous lesions arising from the entodermally-derived digestive system epithelium. The concentration of the tumor antigen was greater in lower bowel malignancies than in those of the upper gastrointestinal tract. It should be emphasized that this antigen was apparently absent from primary cancers in all other tissues, including those of the entodermal portions of the mouth. Benign tumors of the bowel and metastatic lesions of other organs to various parts of the digestive system were also devoid of these components. This would indicate that the presence of the tumor antigen in the cancers in which it was found was dependent upon the tissue of origin, rather than the tissue of growth, of the tumor [14]. The marked localization of the tumor antigen to entodermally-derived digestive organ cancers suggested the possibility of some relationship between this component and the embryonic development of the human digestive system. Hence, a number of embryonic and fetal tissues were examined by gel precipitation. These studies revealed that the antigen, which in the adult was localized to digestive system cancers, could be detected in embryonic and fetal gut, pancreas, and liver in the first two trimesters of gestation [14]. This component was, therefore, named carcinoembryonic antigen (CEA) of the human digestive system. This antigen could not be detected in fetal digestive tissues in the third trimester, or in any other tissue at any time during gestation. The very sensitive techniques which have recently been developed for the detection of CEA suggest that this material may be present in very low concentrations in tissues other than those which have been described above [15,16]. The problem of whether the material found in these tissues is identical to the CEA isolated from digestive system cancers, or if there are "~CEA-like" moieties which may mimic the presence of CEA in sensitive assay systems, will be considered more fully below. IVC. Cellular localization of the CEA [n initial attempts to determine the cellular location of the CEA, unfixed frozen sections of cancerous tissue from the digestive system were treated with fluorescein-

129 conjugated rabbit anti-CEA antiserum. It was observed that staining appeared quite sharply localized to the region of the plasma membrane of the neoplastic cells [17]. Suspensions of viable colon cancer cells were then prepared and incubated with fluorescein-conjugated antiserum. The majority of cells examined by ultraviolet microscopy revealed the pattern of a beaded-necklace or signet ring, typical of immunofluorescent localization of antigens to the surface of the cell. Suspensions of cells from colon cancer in tissue culture were mixed with specific rabbit anti-CEA antiserum. The reaction was followed by phase-contrast microscopy, and revealed a rapid and massive agglutination of cultured cells [18]. Pretreatment of the antiserum with minute quantities of purified CEA completely inhibited agglutination, as it did the immunofluorescent reactions described above. Thus, functionally as well as morphologically, the CEA behaves as a constituent of a plasma membrane or, at least, as a component lying close to the surface of tumor cells. Subsequently, a study was undertaken to assess the position of the antigen in the ultrastructure of the surface of the colon cancer cell. Viable explants of single cells from colon cancers were incubated with ferritin-conjugated goat anti-CEA antiserum. The suspension of cells was then washed thoroughly and processed for examination by electron microscopy. The pattern of localization of the ferritinantibody conjugate revealed that the plasma unit membrane structure itself remained virtually unstained. However, a heavy ferritin label was found in the glycocalyx or "fuzzy coat", immediately adjacent to the surface membrane [19]. Presumably then, CEA is not an integral structural component of the trilaminar image usually referred to as the plasma membrane, but lies even further to the periphery of the cell. It is not certain, however, whether CEA is an integral part of the glycocalyx, or is a secretory product in transit across the cell surface, IVD. Cellular and humoral immune responses to CEA

Both humoral and cell-mediated immunologic reactivity against CEA have been assessed. A specific humoral immune response against CEA, primarily of the IgM class of immunoglobulins, has been demonstrated by at least two techniques [20-22]. It has, further, been observed that the methodology employed for antibody detection is of major importance in determining the results which are ultimately obtained. Hence, bis-diazotized benzidine-hemagglutination produced positive results with the sera of patients bearing digestive system cancers only if the tumor had not undergone metastatic dissemination [20]. On the other hand, the techniques of radioimmunoelectrophoresis and radioimmunochromatography were able to detect anti-CEA antibody in the sera of patients with disseminated cancers [21,22]. The use of a modified Farr technique, even with acid-dissociation for the detection of antibody bound to antigen, was unable to detect anti-CEA antibodies under any circumstances. Hence, it may be that the technologic problems involved, and the reagents employed, have been responsible for the inability of some workers to demonstrate the presence of specific-CEA antibodies in the sera of patients with colon cancer [140,141]. It is

130 noteworthy that such antibodies have been demonstrated in the circulation of pregnant women [20]. A very elegant in vivo demonstration of anti-CEA antibody production in patients with colon cancer, and the potential pathophysiologic effects of the production of such CEA-anti-CEA immune complexes, has recently been reported in a patient suffering from carcinoma of the colon and the nephrotic syndrome. Employing immunofluorescent techniques to examine the kidney biopsy from this patient, it was found that CEA, immunoglobulin, and complement had been deposited in the glomerular basement membrane [23]. Upon examining the area of cell-mediated immune responsiveness to the CEA, it becomes apparent that no substantial evidence for such a phenomenon has ever been put forward. Although cell-mediated immunity against antigens common to human colon cancers and fetal gut epithelium has been demonstrated by the use of the colony inhibition technique, it was simply suggested that the CEA might be the common factor involved, but no studies to investigate this point were performed [24]. Recent work employing delayed skin reactions to intradermally inoculated antigen prepared from various tissues has shown that patients with carcinomas of the colon generally give a positive skin test to extracts of human fetal intestine and colon cancer tissue, but not to similar extracts of normal adult colon. The skin reactive antigen(s) involved, however, appear to be distinct from purified CEA [25,26]. In addition, purified CEA fails to stimulate lymphocyte transformation in tests of patients suffering from colon cancer [27].

IVE. Circulating CEA The initial observation, by bis-diazotized bernidine-hemagglutination, that anti-CEA antibodies were absent from the sera of patients with disseminated bowel cancer, remained to be explained. One possible explanation for this phenomenon was that the CEA in the glycocalyces of the cells of a large tumor mass could serve as an antibody sink, absorbing anti-CEA antibodies as the blood circulated through the neoplastic tissue. A second possibility was that the neoplastic tissue could release CEA directly into the circulation. In the presence of a large mass oftumoT tissue, the corresponding concentration of antigenic material in the serum would be relatively high and could lead to CEA-anti-CEA complexes in antigen excess. Under such circumstances then, the anti-CEA moieties would no longer be available for participation in subsequent serologic reactions in vitro. In order to determine whether or not the CEA could leave the glycocalyx of the tumor cell surface and enter the circulation, a radioimmunoassay for the detection of this material was developed. The technique initially employed was based on the principle of coprecipitation-inhibition in half-saturated ammonium sulfate [28]. The initial study performed demonstrated that the assay could detect nanogram quantities of circulating CEA. The results also suggested that the assay was highly specific for CEA in the sera of patients with bowel cancer. A great deal of work along

131 these lines has since been performed, and the theoretical and practical implications of the results obtained to date will be discussed subsequently.

IVF. Metabolism of CEA The question has been raised as to whether CEA is produced by the tumor in situ, or if this material may be synthesized at another site in the body, and then adsorbed onto the tumor cell surface. Data obtained from studies performed both with tissue-cultured colon cancer cells in vitro [29] and xenografted human colon cancer tissue in hamsters [30] have demonstrated that CEA is produced by tumor cells in situ. There is a great deal yet to be learned about the metabolism of CEA. Studies of CEA concentrations in the sera of patients who have undergone apparently complete bowel tumor resection indicate that the CEA is catabolized rapidly, and serum concentrations postoperatively frequently fall to undetectable levels 2-14 days after surgery [31,32]. The site of CEA breakdown in man remains unknown but animal experiments suggest that the liver is a major site of catabolism of CEA [33]. The serum decay of CEA administered to xenogeneic animals has also been studied, and has shown a rapid disappearance comparable to that seen in the human. In fact, these studies suggested a biologic heterogeneity within any one population of CEA molecules because different rates of decay were observed with time [33].

IVG. Radioimmunoassays for CEA The initial experimental work on CEA, summarized in the foregoing discussion, established the existence of a material associated with carcinomata originating from the entodermally derived portion of the gastrointestinal tract and with certain fetal tissues. It is important to note that the original description of this molecule was through its antigenic activity in the heterologous host. Thus, the tumor site of CEA is defined by antibodies that bind specifically to the molecule but cannot be removed from the antiserum by absorption with normal tissues. Subsequent efforts, in our laboratory and other laboratories, led to a physicochemical description of the molecule which bears this tumor antigenic site. The distinctions between antigenic activity, the tumor site, and the CEA molecule are analogous to the operational descriptions of enzymatic activity, the active site of an enzyme, and the enzyme as a molecular unit, respectively. The method of measurement of CEA activity thus becomes central to the definition of the molecule and, in current practice, the most important method utilized is the radioimmunoassay. The radioimmunoassay for CEA then becomes important in several respects. (1), It is used, as noted above, to define the immunological criteria for CEA; (2), It is used as the method of identification of CEA during isolation, and of demonstration of increase of specific activity with purification; (3), Chemical manipulations of CEA designed to determine the structure of the tumor site are assessed in terms of their effects on the immunologic activity of the molecule as measured by the radioimmunoassay; (4), The radioimmunoassay, because of its

132 ability to measure small amounts of material in serum, is the method of CEA measurement which has been applied to the clinical situation. Therefore, an understanding of the available radioimmunoassays for CEA, together with an appreciation of some of the difficulties of such analyses, is necessary to a proper examination of the physicochemical and clinical data which are presented in subse:luent sections. The assays employed are based on the general principal of "saturation analysis" [34] or inhibition of binding in which material in the test sample (or serum) "competes" with radio-labelled CEA for binding to an anti-CEA anti-serum. The degree of "'displacement" of labelled CEA from the antibody-bound to the non-antibodybound or free fraction reflects the amount of inhibitory material in tile test specimen. This type of assay system can be described in terms of (1) the antigen employed as radiolabelled material and as standard; (2) the antiserum, (3) tile method of separation of antibody bound and free antigen in order to determine the distribution of radiolabelled material in these two compartments, (4) treatment of the test sample before measurement. Antigens used in the various assay systems available are generally purified from hepatic metastases of colonic carcinomata but the method of purification varies with different laboratories (see below for details of purification procedures). There are no universally accepted standards available for use as a reference material, Moreover, immunologic and chemical comparisons are available only for a limited number of preparations of purified antigen. Hence, the identity of the antigens used by different investigator has not been clearly demonstrated and variability in the immunologic activity of the purified standards utilized will affect the results obtained from the assay. The antisera used in the assays are prepared by a variety of immunization schedules in a number of animal species, utilizing different purified preparations of CEA and are absorbed with non-tumor tissue to a variable degree prior to use [28, 35,36]. in our own experience, using purified CEA preparations as the immunogen, the antisera obtained are polyvalent and require absorption with normal serum, normal bowel, and normal lung to be rendered monospecific, it has been postulated [37] that there are at least three types of anti-CEA antisera in use. Some antisera react with the tumor site of CEA, others are directed towards determinant(s) distinct from this site and hence do not distinguish tumor antigen from antigens in normal tissue, while a third group of antisera represent combinations of tumor and normal specifities. The variability that may be introduced by antiserum differences is illustrated by the results of a radioimmunoassay performed on tile urine of patients with bladder cancer utilizing anti-CEA antisera obtained from three different laboratories. The incidence of CEA positivity in this group was 100~,,, 500/o and 0°/,, depending on tile antiserum employed in the assay [38,39]. The radioassays for CEA are most often identified by the procedure used to separate tile antibody-bound from free fractions of the antigen and by whether a preliminary step of perchloric acid extraction of the serum is employed. This initial step is utilized in a number of the assays in order to separate the carbohydrate-rich,

133 perchloric acid-soluble glycoproteins such as CEA, from the perchloric acid-insoluble serum components which may non-specifically interfere with the assay [16,40]. Such a preliminary step lengthens the time required to complete the assay procedure, since prolonged dialysis is necessary to remove the perchloric acid after extraction. Other assay designs (Table VII) have eliminated this pretreatment step and thus shorten the time required to perform the assay to 24-48 hours. The original assay procedure [28] referred to as the "~Farr" or ammonium sulphate technique, utilizes a perchloric acid extraction step and takes advantage of the solubility of CEA in half-saturated ammonium sulphate. Thus, separation of anti-CEA-12s[-CEA from 125I-CEA is achieved by the addition of saturated ammonium sulphate to the incubation mixture. The Z-gel technique [41] also utilizes a perchloric acid extraction step and achieves separation of bound and free antigen by precipitating the anti-CEA-~ZsI-CEA complex with zirconyl phosphate gel. A similar procedure, using zirconyl phosphate gel, but omitting the initial extraction step has also been developed [42]. An alternative technical approach to the problem of separation of bound and free antigen has been the precipitation of the anti-CEA[125I] CEA by a second antibody with specificity towards the anti-CEA antibody [36,43]. For example, a rabbit anti-CEA antibody, together with any bound antigen, can be precipitated by a sheep anti-rabbit IgG antiserum. Two variants of this procedure are in use (Table VII) neither of which employ a step ofperchloric acid extraction. Other less commonly used assay techniques are outlined in Table Vll [44,45]. All the factors, alluded to above, which influence the assay procedure suggest that the inhibition observed in the radioimmunoassay may be of three types [46]: (1), specific, in which the inhibitory molecule contains the tumor site and thus quantitatively competes with CEA standards purified from colonic carcinomas; (2), crossreactive, in which a family of molecules commonly termed "CEA-like", but as yet incompletely defined, compete to varying degrees with colon tumor-derived CEA in the radioimmunoassay; (3), non-specific~ in which the inhibitory material bears no immunologic relationship to CEA [16,40]. Further work will be required in order to define the absolute and relative specificities of the various CEA radioimmunoassays in use, both for experimental and clinical investigations. A recent study sponsored by the National Cancer Institute [47] examined the results of a number of different CEA assays performed on selected sera t¥om both normal individuals and patients with a variety of diseases. It was shown that although a given serum sample yielded different absolute values in different assays, the overall diagnostic concordance between assays was in the range of 85%. Whether the 15 ~i discordant results indicate different specificities of the various assays or represent technical problems is unclear at present. A more detailed analysis of the clinical applicability of the various assays described above is presented in a subsequent section.

134 v. PURIFICATION AND PHYS1COCHEMICAL PROPERTIES OFCEA

VA. Purification of CEA CEA has been purified by a variety of different techniques. In our laboratory, metastatic tumor lesions have been used whenever possible in order to obtain large quantities of cancer tissue from a single source. Purification of the CEA has been achieved by a sequence of procedures including perchloric acid extraction, column chromatography on Sepharose 4B and Sephadex G-200, and preparative block electrophoresis on Sephadex G-25 [48]. Recent modifications in this purification procedure have included removing the perchloric acid by column chromatography on Sephadex G-25 (coarse), and concentration of the eluate by ultrafiltration through an Amicon PM-30 membrane. Alternatively, perchloric acid can be removed by rapid dialysis through an artificial kidney. This eliminates the need for a lengthy dialysis step. In addition, the CEA preparations have been rechromatographed on Sephadex G-200 following block electrophoresis. Other methods of CEA purification have been reported. These have included : lithium diiodosalicylate for CEA extraction and ion exchange chromatography for product separation [49] ; perchloric acid extraction, pevikon electrophoresis, Sephadex G-200 chromatography and isoelectrofocusing [50]; perchloric acid extraction followed by sequential column chromatography on a mixed bed resin of DEAE-cellulose and CM-cellulose, Sepharose 6B, and Sephadex G-200 [51]. Thus, different investigators have employed different techniques for the isolation of CEA. It is still uncertain if the products obtained by the various methods are identical. VB. Physical properties of CEA preparations 1. Molecular size Using molecular sieve chromatography, an approximate molecular weight of 200 000 daltons was obtained for CEA [4&51]. Furthermore, sodium dodecyl sulfatepolyacrylamide electrophoresis, using a mixture of glycoproteins of known molecular weights as standards, suggests a mean molecular weight for CEA of 200000 [52]. Each preparation of CEA obtained in our laboratory from colon cancer tissue has given rise to a single peak upon ultracentrifugation, with a sedimentation coefficient of 7-8 S, which is in keeping with the results of molecular sieve chromatography and sodium dodecylsulfate-polyacrylamide gel electrophoresis [53,54]. However, without the availability of a shape-dependent parameter, partial specific volume, and other physical data, an exact molecular weight cannot as yet be calculated for any given CEA preparation. 2. Molecular cha~L~e CEA has been reported to migrate in the fl-globulin region upon immunoelectrophoresis in agar gel at pH 8.6 [53]. Polyacrylamide gel electrophoresis of purified CEA preparations has yielded a single, but somewhat diffuse, band [52]. Heterogeneity of mobility was particularly notable upon comparison of the products of tumor tissues which had arisen in different sites within the gastrointestinal tract

135 TABLE ! PHYSICAL PROPERTIES OF" CEA Soluble in perchloric acid Soluble in 50~ saturated ammonium sulfate Insoluble in ethanol Heat stable Sedimentation coefficient of 7-8 S p-mobility on agar electrophoresis at pH 8.6 Single polydisperse band on polyacrylamide gel electrophoresis Molecular weight of 200 000 ± 20 000 lsoelectric points of < 3 and 3.75 ± 0.25 [523. Neuraminidase treatment of CEA resulted in a narrow, more homogeneous electrophoretic band, and decreased the variation noted between the different preparations. In our laboratory, isoelectric focusing of purified CEA revealed isoelectric points of 3.0 and 3.';5. However, following neuraminidase treatment of purified CEA, a single homogeneous zone of activity was obtained with an isoelectric point of 5.0 [52]. Other investigators have observed that a characteristic feature of all purified CEA samples which they examined was a double peak of CEA activity in the range of pH 2-3 [55]. They also noted CEA activity with isoelectric points of 3.5, 4.0, 4.25 and 4.5. On the other hand, observations of CEA preparations with single isoelectric points of 4.8 [49] and 4.7 [56] have been reported, lsoelectric focusing of crude tumor extracts yielded peaks of CEA activity at pH 2.4, 3.0, and 4.5 through 6.4 [57].

3. Miscellaneous physical parameters The CEA molecule has been shown to be soluble in water, perchloric acid and half-saturated a m m o n i u m sulphate but is insoluble in ethanol [53]. It is relatively resistant to boiling [58]. The properties of solubility in both perchloric acid and halfsaturated a m m o n i u m sulphate form the basis of a number of the radioimmunoassays for serum CEA presently being employed [28]. (Table 1).

VC. Structure of CEA The carbohydrate/protein ratio of CEA is usually of the order of 3:1. However, a ratio as high as 5 : 1 has been observed in purified CEA of gastric cancer origin and a ratio as low as 1 : 1 has been observed in material obtained from colonic cancer tissue [52].

l. Carbohydrate composition The principle carbohydrate constituent of purified CEA preparations is N-acetylglucosamine which usually constitutes approximately 40 ~ of the total carbohydrate content of each CEA preparation (Table ll). In contrast, N-acetylgalactosamine is either absent or present only in very low concentrations in purified CEA preparations. Some degree of variation is usually noted in the fucose, mannose, galactose and sialic acid contents of preparations of CEA derived from different colonic tumors (Table II) [52].

136 TABLE I1 CARBOHYDRATE COMPOSITION OF CEA Results of carbohydrate analyses of CEA by gas-liquid chromatography obtained in four different laboratories. Samples of CEA were purified from hepatic metastases of colonic carcinomas. The data are expressed as follows: (weight of monosaccharide/weight total carbohydrate of sample) 100~ n.s., no significant amount detected; a, Banjo et al. [52,61, 62], means of results of six different preparations; b, Hansen, H. J. as reported by Kupchik et al. [137]; c, Coligan et al. [63]; d, Westwood, J. [142]. Monosaccharide

a

b

c

d

Fucose Mannose Galactose Sialic acid N-acetylgalactosamine N-acetylglucosamine

I 1.3 12.8 19.9 8.1 2.6 43.3

12.0 21.7 15.6 7.9 n.s. 42.3

15.2 14.3 17.9 10.7 1.8 40.2

21.0 I 1.3 26.6 3.9 I. 1 35.7

Characterization, by paper chromatography, of the sialic acid liberated from C E A by mild acid hydrolysis, indicates that it is N-acetyl-neuraminic acid [59]. The sialic acid moiety is usually the most variable of the carbohydrate constituents [52, 53]. Thus, it has been shown that two groups of C E A molecules, which show a 5-fold difference in sialic acid content, can be separated on a mixed-bed, ion exchange resin [35]. Similar results have been obtained by epichlorhydrin triethanolamine-cellulose c h r o m a t o g r a p h y and by iso-electrofocusing [54,59]. The variability in sialic acid concentration may well explain a good deal of the observed electrophoretic heterogeneity within, and between, C E A preparations [65]. In addition, sialic acid variability may be the basis for the observation of at least two populations of biologically distinguishable C E A moieties which have been detected during studies of plasma C E A disappearance in xenogeneic animals mentioned previously [33].

2. Structure of the carbohydrate moiety Glycoproteins are complex macromolecules consisting of a variable number of oligosaccharide subunits and one or more polypeptide chains. Those glycoproteins studied to date have revealed structures consisting of linear and/or branched oligosaccharide chains of variable length attached to the polypeptide chains via O-glycosidic linkages to serine or threonine, or via a peptide bond to asparagine. A n u m b e r of different analytic approaches can be used to elucidate this type of structure. a. Acid hydrolysis of CEA. Partial acid hydrolytic cleavage of oligosaccharide chains to obtain low molecular weight fragments is an established technique for the elucidation of the structure of glycoproteins [60]. The release of monosaccharides or oligosaccharides at defined time intervals provides information on the sequence of the various components within the chain. In addition, the isolation and subsequent characterization of such fragments may indicate the mode o f linkage between the individual sugars.

137 On the basis of the foregoing rationale, purified CEA was subjected to controlled hydrolysis with polystyrene sulphonic acid at pH 2.4 at 60 °C and 86 ° C [61]. Heterosaccharide fragments were released which contained either mannose and N-acetylglucosamine or N-acetylglucosamine alone. Further structure elucidation was not possible since only trace quantities of the purified fragments were obtained. b. Enzymatic hydrolysis ofCEA. Although numerous glycosidases are available which can degrade isolated glycopeptide fragments, only a few such enzymes will degrade the native intact molecule. However, these latter enzymes are not readily available in a highly purified form and their enzymatic specificities have not been well characterized. The action of neuraminidase on sialic acid-containing glycoproteins has been intensively studied in the last decade. A preparation of neuraminidase from Clostridium perfringens was used to remove the sialic acid from CEA [52,62]. Similar experiments were performed with neuraminidase from Vibrio eholerae [63,64] and both enzymes were able to remove 100 To of the sialic acid residues present in CEA, demonstrating the presence of sialic acid in terminal positions in the CEA molecule. Attempts to remove galactose residues from CEA by fl-galactosidase have been unsuccessful [64]. Oxidation of CEA with galactose oxidase with subsequent reduction with NaB3H4 indicates the presence of at least 2 residues of galactose at terminal positions, and 10-12 galactose residues in penultimate positions of the putative carbohydrate branches of the CEA molecule [62,65]. Attempts to remove sugars other than sialic acid from CEA with exoglycosidases have not been successful [64]. Nevertheless, the enzymatic approach, using combinations of appropriately purified endoglycosidases and exoglycosidases to elucidate the carbohydrate structure of CEA, should be pursued with vigor. c. Periodate oxidation of CEA. Periodic acid and its salts, which cleave carboncarbon bonds between adjacent dihydroxy positions in sugar moieties of glycoproteins, have been most useful in the structural analyses of such molecules. Periodate oxidation of CEA has resulted in complete destruction of sialic acid and fucose, the elimination of 25-50~o of mannose and galactose, but no destruction of N-acetylglucosamine. The CEA preparations so treated showed no loss of CEA antigenic activity [63,143]. It is important to note, however, that since reduction and hydrolysis were not carried out, the hemialdal groups or the aldehyde groups of the sugars were still attached to the oxidized molecule. These modified groups may still be immunologically active despite the absence of the precursor monosaccharides from the sugar analysis. d. Carbohydrate components of glycopeptides. Following neuraminadase treatment, the resultant sialic acid-free CEA was exposed to the non-specific protease nagase. Six glycopeptides were isolated by sequential chromatography on Sephadex G-25, cellulose powder and the cation exchange resin A G 50W-X4. On the average, these glycopeptides were 92 ~ carbohydrate in composition. The molecular weights of these glycopeptides were estimated to be of the order of 4000 by Sephadex G-25 chromatography. This would suggest that the CEA molecule may be composed of

138 TABLE II1 AMINO ACID COMPOSITION OF CEA Results of amino acid analyses of CEA obtained in four different laboratories. Samples of CEA were purified from hepatic metastases of colonic carcinomas. The data are expressed as follows: (weight of amino acid/weight total protein of sample) • 100~. n.d., not done; a, Banjo et al. [52, 61, 62], means of results of six different preparations; b, Hansen, H. J. as reported by Kupchik et al. [137]; c, Coligan et al. (55), Terry et al. [67]; d, Westwood, J. [143]. • In a, b and c, determinations of cysteic acid were not performed. Hence, cysteine converted to the cysteic acid form during hydrolysis would not have been detected. In d, the content of cysteine was determined as cysteic acid after oxidation of the molecule by performic acid [66]. Amino Acid

a

b

c

Aspartic acid Glutarnic acid Serine Threonine Isoleucine Leucine Proline Glycine Alanine Valine Tyrosine Phenylalanine Lysine Histidine Arginine Cysteine* Methionine Tryptophan

14. I 11.3 8.4 8.0 6.0 10.3 6.5 3.3 3.8 5.8 3.6 3.3 2.9 2.0 4.6 0.0 0.0 n.d.

16.6 12.4 10.5 8.6 5.8 9.4 4.6 4.2 3.9 6.6 5.8 2.7 2.9 1.8 4.2 0.0 0.0 n.d.

15.0 11.6 8.1 7.9 5.0 9.0 10.0 3.| 4. I 6.3 5.6 3.8 3.4 2.4 4.9 0.0 n.d. n.d.

d 14.7 10.6 10.4 9.6 4.7 8.2 8.4 5.5 6.2 7.3 3.6 2.2 2.8 1.8 3.3 0.8 trace n.d.

side chains c o n t a i n i n g approximately 17 sugar residues. F u r t h e r m o r e , the variable m o n o s a c c h a r i d e c o n t e n t of these glycopeptides suggests that the c a r b o h y d r a t e s u b u n i t s are heterogeneous i n nature [62].

3. Amino acid composition of CEA. A p p r o x i m a t e l y 5 0 ~ of the protein portion of C E A is composed of a m i n o acids with a terminal carboxyl a n d hydroxyl group in their side chains (Table III). The other a m i n o acid residues are usually present in lower proportions and the s u l f u r - c o n t a i n i n g a m i n o acid m e t h i o n i n e is usually absent. The presence of 12 residues of cysteine in the C E A molecule has been reported [66]. Although a significant degree of variability may be observed in the a m i n o acid c o m p o s i t i o n of different colonic C E A p r e p a r a t i o n s when the a m i n o acid c o n t e n t is expressed o n the basis of total weight of material, the a p p a r e n t variability in a m i n o acid c o n t e n t of such C E A preparations is reduced when the c o n t e n t of any a m i n o acid residue is calculated on the basis of the total protein p o r t i o n of the molecule [52]. A c o m p a r i s o n of six different colonic C E A p r e p a r a t i o n s from our l a b o r a t o r y with three p r e p a r a t i o n s from three other laboratories showed similar results with respect to most of the a m i n o acid residues (Table III). However, a degree of variability

139 was observed in the chemical composition of CEA derived from a gastric tumor when compared to colonic cancer preparations [52].

4. Primary structure of the protein moiety of CEA. The N-terminal amino acid in CEA purified from liver metastases of colon cancer has apparently been shown to be lysine by different investigators [67,68,90]. The identity of the C-terminal amino acid has not been established as yet. Amino acid sequencing of five different CEA preparations from colonic tumors shows identical sequences of the first 20 to 30 amino acid residues [67,68]. Similar sequences were obtained upon analysis of CEA produced by a human colonic tumor line maintained in hamsters and from CEA isolated from human serum [59]. However, quantitative analyses and recoveries of the residues cleaved at each position have not been reported. Thus, the possibility that the protein backbone is heterogeneous has not been ruled out. VD. Nature of the carbohydrate-protein linkage in CEA The elucidation of the carbohydrate-protein linkage(s) in the CEA molecule requires a knowledge of both the amino acid and monosaccharide residues which are involved, and the nature and position of the functional groups forming the bond(s). The most commonly described carbohydrate-protein linkages involve the amino sugars. The glycosylamine linkage of some glycoproteins has been found to involve the amide group of asparagine and the glycosyl group of N-acetylglucosamine [69]. A second type of bond involves N-acetylgalactosamine linked, O-glycosidically, to serine or threonine residues [70]. Studies of the glycopeptides obtained from CEA have provided indirect evidence for the involvement of N-acetylglucosamine in the carbohydrate-protein linkage [62]. It is also of interest that asparagine-N-acetylglucosamine, obtained from Dr F. Malley, displays weak inhibitory activity in the CEA radioimmunoassay [71]. The absence or the low content of N-acetylgalactosamine in CEA, and the inability to remove heterosaccharides from CEA upon mild alkaline treatment, make it unlikely that N-acetylgalactosamine is involved in the carbohydrate-protein linkage [64,65]. VE. Immunochemical studies of CEA 1. Binding studies between 125I-CEA and various antisera Antisera against CEA have been prepared in heterologous animals, including rabbits, goats, and horses. In order to render these sera tumor-specific, they have been absorbed with an excess of appropriate normal tissue extracts [52]. These anti-CEA antiserum preparations have been used to establish radioimmunoassays (v.i.) for the measurement of CEA activity in human blood and tissues, and in chemically or enzymatically derived fragments of the native CEA molecule. In addition to antibodies directed against the tumor-specific site, the sera of cancer patients and pregnant women also bind to CEA [20,21]. It is unknown if these homologous antibodies bind to the same site as anti-CEA raised in heterologous species. Anti-blood group substance antibodies have been reported to bind CEA (Table

140 TABLE IV BINDING STUDIES OF CEA The ability of various antisera and lectins to bind to CEA or IzsI-CEA was examined by primary binding techniques (radioimmunoassay) or by precipitation in agar gels. ~ indicates that significant binding was demonstrable; - - indicates that significant binding could not be demonstrated. Material Tested Antiserum to CEA (heterologous antiserum) CEA (human antiserum) Blood group A Blood group B Blood group Le~ Blood group 1 Blood group i Pneumococcus type xiv Fetal sulfoglycoprotein Lectins Concanavalin A Wheat germ agglutinin Phytohemagglutinin Ricin

Result

References

I t E -i ~ ----~

[14] [20, 21 ] [72] [74] [74] [64] [64] [64] [75]

+ + ? :

[83, 91] [91 ] [64] [67]

IV). A n t i - b l o o d g r o u p A a n t i s e r u m was c a p a b l e o f b i n d i n g up to 11 ~ of ~2sI-CEA [72] despite the absence, in m a n y instances, o f N-acetylgalactosamine, the i m m u n o d o m i n a n t sugar o f the b l o o d g r o u p A substance. It is o f interest in this context t h a t a n t i - A a n t i b o d i e s react with placental c h o r i o n i c g o n a d o t r o p h i n even t h o u g h Na c e t y l g a l a c t o s a m i n e is a b s e n t from this molecule [73]. Others have shown t h a t antib l o o d g r o u p B a n d Lewis-type a n t i b o d i e s react with the C E A molecule [74]. H o w ever, anti-I, anti-i a n d horse a n t i - p n e u m o c o c c u s type X I V antisera d o not b i n d to purified C E A [64]. A n t i s e r u m p r o d u c e d against fetal s u l f o g l y c o p r o t e i n antigen, a molecule f o u n d in gastric juice o f patients with s t o m a c h cancer, was c a p a b l e o f b i n d i n g to C E A . This suggests the presence o f shared antigenic sites with the fetal sulfoglycoprotein m o l e c u l e [75]. A n t i s e r a raised in h e t e r o l o g o u s species to a 60 000 m o l e c u l a r weight constituent f o u n d in n o r m a l bowel and lung a n d variously called non-specific cross reacting antigen colonic c a r c i n o m a antigen, etc., react with C E A (see section on crossreactive material).

2. Specific leetins Several p l a n t agglutinins have been shown to agglutinate t u m o r cells a n d t r a n s f o r m e d cells m o r e effectively t h a n cells derived f r o m the n o r m a l tissue [76,77]. Since C E A is a g l y c o p r o t e i n f o u n d on the surface o f t u m o r cells, the ability o f several lectins o f defined specificity to b i n d to C E A has been studied (Table IV). T h e lectins

141 employed were c o n c a n a v a l i n A, wheat germ agglutinin, p h y t o h e m a g g l u t i n i n , a n d ricin.

C o n c a n a v a l i n A binds polysaccharides c o n t a i n i n g multiple t e r m i n a l n o n -

reducing a-D-glucopyranosyl, a - D - m a n n o p y r a n o s y l , fl-D-fructofuranosyl or a-Da r a b i n o f u r a n o s y l residues [78]. W h e a t g e r m agglutinin reacts specifically with terminal n o n - r e d u c i n g N-acetylglucosamine a n d its disaccharide, di-N-acetylchitobiose [76], a n d to a lesser extent with N - a c e t y l - n e u r a m i n i c acid [79]. P h y t o h e m a g g l u t i n i n binds to more complex heterosaccharides such as that f o u n d in fetuin [80], while ricin binds D-galactose [81 ]. I m m u n o d i f f u s i o n in agar gel was used to study the interaction between C E A a n d lectins. T o q u a n t i t a t e the reactions of lectins with CEA, a modification of the r a d i o i m m u n o a s s a y for C E A was used. C E A was shown to b i n d to all four lectins to variable degrees as shown by reactions in agar gel a n d interaction with 1251-CEA [64,82,83]. Thus, in lectin excess c o n c a n a v a l i n A was able to b i n d a m a x i m u m of 50 of 125I--CEA. This b i n d i n g was specifically inhibited by a - D - m a n n o p y r a n o s y l a n d N-acetylglucosamine. The b i n d i n g of wheat germ agglutinin to 9 0 ~ of 125I-CEA, was inhibited by N-acetylglucosamine. P h y t o h e m o g g l u t i n i n b o u n d a m a x i m u m of

TABLE V INHIBITION STUDIES OF CEA-ANTI CEA SYSTEM The following materials were tested for their capacity to cause inhibition of binding of 12sI-CEA by an absorbed anti-CEA antiserum. The relative inhibitory weight is derived as follows: (weight of tested material required to achieve 50 ~ inhibition of binding)/(weight of purified CEA required to achieve 50~ inhibition of binding). * indicates that no significant inhibition of binding was demonstrable. Materials

Relative

References

Low molecular weight materials (see Table VI) Monosaccharides Amino Acids Oligosaccharides Glycopeptides IgM glycopeptide Asparagine-N-acetylglucosaminet CEA-derived glycopeptides Macromolecules CEA Blood group A Blood group B Blood group Le a Blood group H Pneumococcus type xiv substance Gastric mucosa A substance Streptococcal group A antigens ~-acid glycoprotein Ovomucoid Ovarian cyst fluid material

inhibitory weight * * *

[91] [91 ] [91 ]

* 1.2 • 106 10--330

[64] [71 ] [62]

1 3.3 • 103, 1.3 • 103 2.6 • 103 73.3 • 103 1 • 10a * 4.6 " 105 * * * 1 • 103

[62] [59, 72] [74] [74] [74] [64,74] [59] [64] [64] [64] [62,136]

t 2-acetamido- 1-N-(4"-L-aspartyl)-2-deoxy-~-D-glucosylamine

142 TABLE VI MATERIALS DEVOID OF INHIBITORY ACTIVITY The following materials were tested for their capacity to cause inhibition of binding of ~251-CEA by an absorbed anti-CEA antiserum. Each material was tested in amounts ranging up to 10mg or 10~zM, and in no case was any significant inhibitory activity demonstrable. L-fucose,D-arabinose, D-mannose, D-galactose, o-glucose, D-glucosamine, D-galactosamine, D-mannosamine, N-acetyl-D-glucosamine, N-acetyl-t~mannosamine, 2-deoxy-D-glucose, s-methyl-D-glucose, 7-methyl-D-galactose, c~-methyl-D-mannose,[5-methyl-N-acetyl-o-galactosamine. [3-D-gentibiose,o-raffinose, C~-D-melibiose, stachyose, ~-L-fUCO-(I-2)-~-Ogal-(l-4)-D-glc(2-fucosyllactose), C~-L-fUCO-(1--2)-t3-D-gaI-(1--4)-{~-L-fUCO(1--3)} -D-glc (lactodifuco-tetraose), ~-o-gal-(1-3)-~-o-glcNac(l-3)-[3-o-gal (1-4)-D-glc(lacto-N-tetraose) ~-L-fUCO-(1-2)-[3-D-gaI-(I-3)-~-D-glcNac-(I3)~-D-(1-4)-D-glc(lacto-N-fucopentaose I), ~-D-gal-(1-3)-{ ~-L-fUCO-(14) } [3-D-glcNac-(l-3) [3-D-gal-(1-4)-D-glc(lacto-N-fucopentaosell), ~3-D-glcNac(l-3)-D-gaI-Er, di-N-acetylchitobiose, chitobiose, (N-acetyl-o-glucosamine)3 (mannose)~,. 20 common anaino acids

Monosaccharides

Oligosaecharides

Amhlo Acids .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

80% of 125-CEA. However, in contrast to the other lectins, the binding of phytohemagglutinin to CEA was not inhibited by monosaccharides such as D-galactose, N-acetylglucosamine or D-mannose. The binding of ricin to CEA has been reported but quantitative studies were not done [59]. 3. Comparative inhibition studies in the 125I-CEA-Anti-CEA system Using the radioimmunoassay for CEA, the ability of various macromolecules, oligosaccharides, monosaccharides, and amino acids to inhibit the binding of antiCEA antisera to l z s I - - C E A was assessed. The results, shown in Table V, revealed that, on a weight basis, 1000-3300 times more blood group substances, compared to CEA, are required to produce 50~,i inhibition of 12Sl-anti-CEA binding. Neither pneumococcal polysaccharide type X1V, a precursor of ABH blood group antigens, nor the carbohydrate moiety of group A streptococci, which contains 11 N-acetylglucosamine residues in terminal positions produced any inhibition. Similar results were obtained with a-acid-glycoprotein and ovomucoid. Inhibition studies with the monosaccharides, amino acids, and oligosaccharides shown in Table V1 showed no inhibition of 125I-CEA-anti-CEA binding. 1.2" l 0 6 times more asparagine-N-acetylglucosamine than CEA, on a weight basis, was required to achieve 50~o inhibition of the CEA-anti-CEA reaction. However, glycopeptides derived from nagase-digested CEA were relatively good inhibitors, requiring 10-330-fold more material, by weight, to achieve comparable inhibition to intact CEA [62]. 4. Materials" cross-reacting with CEA A series of materials has been identified in various normal and cancerous tissues which shows partial identity with CEA in agar gel diffusion against unabsorbed

143 anti-CEA antisera. These substances have been variously termed: non-specific cross reacting antigen [84], normal glycoprotein [85], colonic carcinoma antigen III [51], breast carcinoma glycoprotein [86], and membrane associated, tissular autoantigen [87]. With the exception of breast carcinoma glycoprotein, which has a molecular weight of 200 000, most of these materials have a molecular weight of about 60 000, and exhibit a fl-mobility on agar gel. It appears that normal glycoprotein, CEX, [88], and non-specific cross reacting antigen given reactions of identity in agar gel and are in turn partially identical to fetal sulfo-glycoprotein antigen [89]. It is not certain if colonic carcinoma antigen, CCEA-2 [90] and breast carcinoma glycoprotein are identical to the other cross-reacting substances of fl-mobility noted above. More importantly, except for CCEA-2 none of these materials has been sufficiently characterized to allow direct comparison with CEA. Their biologic relationship to CEA is unknown and it is as yet uncertain whether they are responsible for some or all of the "CEA activity" detected by radioimmunoassay in various tissues and sera noted previously.

VF. The tumor-specific grouping of the CEA molecule The results of studies described earlier [61] using polystyrene sulphonic acid for the partial hydrolysis of CEA strongly suggested that a heterosaccharide grouping consisting solely, or largely, of N-acetylglucosamine was of major importance in the tumor-specific antigenic site of the CEA molecule [61]. However, the monosaccharide N-acetylglucosamine was not able to inhibit ~zsI-CEA-anti-CEA binding [91]. Nevertheless, the central role played by this amino sugar was further illustrated when the proteolytic enzyme nagase was used to prepare a variety of immunologically active glycopeptides from the CEA molecule [62]. These fragments were of various molecular weights, between 1000 and 5000. Different fragments retained tumorspecific antigenic activity in the absence of one or more of the monosaccharides, sialic acid, mannose, galactose, fucose and N-acetylgalactosamine. In all instances, N-acetylglucosamine was present in relatively high concentrations in the small immunologically-active fragments. Studies of the intact CEA molecule have revealed that upon removal of all the sialic acid residues of CEA by neuraminidase, no loss of activity results [52,55,62]. Furthermore, upon periodate oxidation of CEA, all of the sialic acid and fucose, and 20-25 ~ of the mannose and the galactose residues are destroyed without any decrease in the inhibitory activity of the CEA molecule [63, 143]. It is worth noting that there was, however, no destruction of any N-acetylglucosamine residues during periodate oxidation. The glycopeptide fractions obtained by the nagase degradation of the CEA molecule each contained 7 or 8 amino acid residues. In every instance, aspartic acid or asparagine and glutamic acid or glutamine were the major amino acid constituents. Hence, it may well be that either one or both of these residues may be covalently bonded to the carbohydrate moiety of CEA. Recently, evidence has been obtained in our laboratory that a molecule consisting of aspartic acid bound to N-acetylglucosa-

Preliminary extraction

1.0 M perchloric acid 1.0 M perchloric acid

none none none none none

Method

Indirect Thomson et al. [28] Hansen et al. [41]

Direct Egan et al. [36] McPherson et al. [44] Coller et al. [45] MacSween et al. [43] Go et al. [42]

second antibody solid phase radioimmunoelectrophoresis second antibody zirconyl phosphate gel

half-saturated ammonium sulfate zirconyl phosphate gel

Method of separation of antibody-bound CEA from free CEA

RADIOIMMUNOASSAYS FOR THE M E A S U R E M E N T OF CIRCULATING CEA

TABLE VII

2 5 0.083 2 2

Time to complete assay (days)

16 2.5 not quantitative 5.0 2.0

2.5 2.5

Upper limit of normal (ng/ml)

4.~

145 mine via an N-acetylglycosylamine-type linkage is capable of weak inhibition of lzsI-CEA-anti-CEA binding (Table V) [71]. This observation suggests that the immunodominant grouping of CEA may not be present at the terminal non-reducing end of the complete heterosaccharide chain of CEA, as it is in other glycoprotein and polysaccharide antigens [92]. It is still, however, possible that the determinant group is present in the terminal position of an incomplete heterosaccharide chain. Although the results strongly suggest that the carbohydrate moiety is intimately involved in the immunodominant tumor site of the CEA molecule, the protein moiety may make an important contribution to the conformation of the tumor-specific site. Recent data [66] indicate that reduction and alkylation of the CEA molecule may markedly diminish its binding activity to anti-CEA. This would support the possibility that protein conformation is of some importance in the binding activity of the CEA determinant.

VG. Nature of the circulating substance(s) measured by radioimmunoassay To date, the nature of the circulating substance(s) measured by the CEA radioimmunoassay in the sera of normal individuals and of patients with various diseases has not been defined. It has not been possible to adequately purify the "CEA-active" material from sera in sufficient amounts to allow immunologic and physicochemical comparison with colon cancer-derived CEA. Thus, it has not as yet been demonstrated that CEA present in the circulation of patients with colon cancers is identical to that purified from the primary or secondary tumors. Moreover, there are insufficient data regarding the CEA-active materials measured by radioimmunoassay in the sera and tissues of some patients with diseases other than gastro-intestinal carcinomas. Some preliminary work along these lines has been reported. For example, extraction of the CEA-active material from various non-malignant tissues has yielded from 1/20 to 1/5000 of the amount found in colonic tumors (93-95). There has, however, been no demonstration that these materials show increasing specific activity with purification. It is of interest, in this context, that nude mice implanted with human colonic tumors develop seropositivity for CEA while CEA activity is absent from the circulation of such mice implanted with human breast or lung tumors [96]. Finally, it has been demonstrated that the CEA-positivity of 15 of 17 sera from patients with ulcerative colitis was abolished by extraction with perchloric acid [97]. These data suggest the possibility that the various "CEA-active" and "CEAlike" materials measured by radioimmunoassay may not be identical to colon cancer CEA and that further work will be necessary both with regard to purification and analysis of these materials and with respect to defining specificity of the radioimmunoassay. Vl. CLINICAL APPLICATIONS The application of quantitative assays for CEA to clinical medicine may be regarded as clinical "spin-off" from studies directed toward the investigation of a

146 TABLE VIII CIRCULATING CEA IN VARIOUS CLINICAL STATES Data are expressed as percentage of subjects within given category with abnormal levels of circulating CEA. The percentages indicated were chosen as being representative of the majority of the clinical data published, and in most instances, are derived from the largest series available. It should be noted, however, that a wide range of positivity has been reported for many of the clinical states mentioned. Clinical status Normal Healthy, unselected Smokers Non-smokers Malignant Diseases Colorectum Stage A Stage B Stage C Stage D Stomach Pancreas Liver Lung Breast Bladder Gynecologic Prostatic Lymphoid Non-malignant diseases Cirrhosis of liver Ulcerative colitis Chronic lung disease Pancreatitis Diverticulitis

Percent positive

Reference

11 19 3

[99] [99 ] [99]

83 45 54 71 89 61 92 63 77 47 42 65 40 36

[32] [108] [109] [108] [108] [99] [99] [I 15] [127] [108 ] [134] [121 ] [133] [99]

45 32 57 43 12

[105] [99] [99] [I 35 ] [ 108]

basic p r o b l e m in h u m a n t u m o r i m m u n o b i o l o g y . T h e utilization o f the initial procedure in clinical studies suggested that the detection o f circulating C E A was virtually d i a g n o s t i c for bowel cancer [28]. However, the d a t a o b t a i n e d f r o m n u m e r o u s subsequent investigations have required m o d i f i c a t i o n o f this early view [31,32,41,42,98 108]. S o m e o f these recent d a t a are outlined in T a b l e V I I I which lists the percentages o f positive C E A tests in n o r m a l subjects and in patients with a variety o f conditions. T h e percentages i n d i c a t e d were chosen as being representative o f the m a j o r i t y o f the clinical d a t a published, a n d in m o s t instances, are derived f r o m the largest series available. It should be noted, however, t h a t a wide range o f positivity has been r e p o r t e d for m a n y o f the clinical states listed. The cut-off points for positivity are those designated by the i n d i v i d u a l a u t h o r s and no a t t e m p t has been m a d e to subdivide the results on the basis o f the type o f assay e m p l o y e d . As indicated previously, the c o n c o r d a n c e between the various assays is high, and, at least with regard to clinical

147 applications, there is no evidence that the available assays differ greatly in their overall sensitivities. It should be noted that the term " C E A " as applied to the clinical situation refers to immunologic activity as measured by a radioimmunoassay. This "immunoassayable" activity may be due to the CEA as it has been defined physico-chemically to date, or to materials that cross-react with CEA and await physico-chemical definition. At present, the ultimate role of the CEA assay in the practice of medicine remains to be completely defined. However, there are sufficient data available to indicate the clinical situations in which the assay is useful and to suggest areas which warrant further investigation. VIA. Diagnosis of cancer The assay in its present form should not be employed indiscriminately as a tool for screening apparently healthy individuals in the population because negative tests may be obtained in subjects with early carcinomas [99,104,108-1 I0]. On the other hand, there is evidence from our clinic which suggests that the study of selected groups of individuals may be of predictive diagnostic value [111]. In this study, 47~, of 81 patients with non-specific gastrointestinal complaints and positive CEA assays were found to harbor gastrointestinal malignancies during longitudinal surveillance with repeated radiologic and endoscopic investigations over a two-year period. The assay may thus be useful as a screening tool in a limited sense; e.g., patients admitted to hospital, those attending a gastrointestinal clinic, or patients with non-specific gastrointestinal complaints not readily attributable to a specific cause. A positive CEA assay does not establish the presence of a malignancy and it cannot be used by itself without recourse to other clinical and laboratory data. There are several studies which have compared the diagnostic value of CEA with other investigative procedures. This type of analysis of the assay with respect to pancreatic carcinoma demonstrated that the CEA assay was more frequently positive than any other diagnostic test used, including upper gastrointestinal barium series, hypotonic duodenography, coeliac arteriography, and percutaneous transhepatic cholangiography [112]. This study, however, did not clearly demonstrate that the addition of the assay to the investigative procedures led to the detection of malignancies which would otherwise have been missed, or altered the clinical outcome. There is evidence to suggest that the detection rate of colonic cancers is highest when the CEA assay and barium enema are used in conjunction than when either test is used alone [113]. The impact of the "false-positive'" assays is different when the test is used adjunctively for diagnosis than when it is used for "single-sample" screening. First, a number of the diseases which manifest "false-positive" results are readily distinguishable from carcinoma (severe alcoholic cirrhosis, emphysema) and can be excluded clinically when a specific individual rather than a population is being examined. More important, many of the non-malignant conditions associated with positive assays show transient rather than persistent elevations and are distinguishable from carcinoma on this basis [99,101,114.]. Finally, most of the false-positive levels tend to

148 be lower than those found in malignant conditions [41~101-105, 115]. Thus, it is not entirely useful to simply compare the raw incidence of positivity in various diseases. A definitive conclusion regarding the diagnostic usefulness of the test will require a prospective study with long-term surveillance, serial sampling, a comparison of tke assay with other diagnostic modalities, and an examination of whether use of the assay will lead to the discovery of otherwise undetected malignancies and/or affect tile prognosis through earlier diagnosis.

VIB. Use of assay for the surveillance of cancer patients More definitive data is available to indicate that the assay is useful in the management of patients whose diagnosis has already been established. In general, in patients with carcinomas of the colon and rectum, stomach, pancreas, bronchus, breast and bladder, very high CEA levels, usually more than 20 ng/ml, indicate the spread of tumor beyond the original site and thus suggest a poorer prognosis [32,41, 99,102,104,107-109,111,117-121,127]. It is important to note that a negative test in such a patient does not rule out the presence of metastatic disease, but usually suggests resectability of the lesion [31,109,116,117,121]. The postoperative decline of elevated CEA levels to the normal range shows good correlation with apparent complete resection of the tumor and potential surgical "cure". This has been demonstrated in patients with gastrointestinal [31,32,99,102, 108,109,122 124], m a m m a r y [108,117], bronchogenic [118,127], urothelial [128], and gynecologic [125,126] cancers. Further, the majority of patients studied over longterm periods (up to 3 years) who maintain negative postoperative levels do not show any evidence of recurrence [102,109,117,127,130]. On the other hand, the failure of a CEA level to drop to normal during the postoperative follow-up period is usually associated with incomplete tumor resections [32,102,109,116,123,124,128], and a steadily rising, rather than transient, increase in the CEA level post-resection indicates progression of the disease [99,109,117,118,123,127,129,130]. More significantly, the reversion to positive of a previously negative test has been shown to be a sensitive and reliable indicator of recurrence of disease [32,108,109,123,124,126,130]. The time interval between "'CEA recurrence" and appearance of clinical or other laboratory evidence of tumor recurrence has varied from two weeks to ten months [32,116,118, 123,125-127,130]. Serial CEA determinations may therefore allow initiation of antitumor therapy at an earlier stage of progression of the disease. This approach becomes especially interesting with the advent of immunotherapy and the notion that such intervention, to be successful, must be initiated when the tumor load is very small; i.e., at clinically undetectable levels. Another area requiring further exploration has been suggested by studies showing that CEA levels correlated well with regression or progression during chemotherapy of carcinomas of the gastrointestinal tract [116,129,131,132], lung [I 27,131] and breast [117] and that the test may be useful to monitor the effectiveness of such therapy.

149 VII. FUTURE PERSPECTIVES A n u m b e r of problems await resolution. F o r e m o s t a m o n g these are the chemical structure of the tumor-associated site; the relationships between the molecule originally described a n d its cross-reacting relatives; the biologic role of C E A in cellular homeostasis; its relationship to the process of m a l i g n a n t t r a n s f o r m a t i o n ; and the clarification of the role of C E A assays in the clinical a r m a m e n t a r i u m . It is likely that the tumor-specificity of the C E A resides in a restricted part of the molecular substructure. The characterization of this region will provide the specific tools required for a n experimental a p p r o a c h to the resolution of the other problems m e n t i o n e d previously. Investigations of C E A d u r i n g the past decade have focused on the description of the p h e n o m e n o n of tumor-specific antigenicity in man, the characterization of the molecule, a n d the application of this knowledge in the clinical setting. The c o n t i n u e d e x a m i n a t i o n of C E A as a physicochemical entity and as a m e m b e r of a particular class of molecules will further the two aims u n d e r l y i n g any efforts in oncology, namely, the i m p r o v e m e n t of patient care and a better u n d e r s t a n d i n g of the basic nature of the neoplastic process.

ACKNOWLEDGEMENTS The authors wish to acknowledge the generous financial assistance of the N a t i o n a l Cancer Institute of C a n a d a , T o r o n t o , the Medical Research Council of C a n a d a , Ottawa a n d the Cancer Research Society, Inc., Montreal, C a n a d a .

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Snell, G. D. (1975) Annu. Rev. Microbiol. 1I, 439458 Old, L. J. and Boyse, E. A. (1965) Fed. Proc. 24, 1009-1017 Prehn, R. T. (1965) Fed. Proc. 24, 1018-1022 Oettgen, H. F. and Hellstrom. K. E. (1973) Cancer Medicine (Holland, J. F. and Frei, E., eds), p. 951-990, Lea and Febiger, Philadelphia Hellstrom, K. E. and Hellstrom, I. (1970) Annu. Rev. Microbiol. 24, ~}73-398 Hellstrom, K. E. and Hellstrom, 1. (1973) Adv. Immunol. 18, 209 277 Baldwin, R. W., Embleton, M. J., Price, M. R. and Robins, A. (1974) Cancer Res. 34, 1452-1460 Bansal, S. C. and Sjogren, H. O. (1972) Int. J. Cancer 9, 490-509 Smith, R. T. (1968) New Engl. J. Med. 278, 1207-1214 Herberman, R. B. (1973) lsr. J. Med. Sci. 9, 300 307 Halterman, R. H., Leventhal, B. G. and Mann, D. L. (1972) New Engl. J. Med. 278, 1272 1274 Order, S. E., Chism, S. E. and Hellman, S. (1973) Natl. Cancer Inst. Monogr. 36, 139-145 Gold, P. and Freedman, S. O. (1965) J. Exp. Med. 121,439-462 Gold, P. and Freedman, S. O. (1965) J. Exp. Med. 122, 467-481 Martin, F., Martin, M. S., Bordes, M. and Bourgeaux, C. (1972) Eur. J. Cancer 8, 315 321 Pusztaszeri, G. and Mach, J. P. (1973) Immunochemistry 10, 197-204 Gold, P. (1970) in Carcinoma of the Colon (Burdette, W. J., ed.), pp. 131-142, Thomas, Springfield, I11.

150 18 !9 20 21 22 23

Gold, P., Gold, M. and Freedman, S. O. (1968) Cancer Res. 28, 1331-1334 Gold, P., Krupey, J. and Ansari, H. (1970) d. Natl. Cancer Inst. 45, 219-222 Gold, P. (1967) Cancer 20, 1663 1667 Gold, J. M., Free:tman, S. O. and Gold, P. (1973) Nat. New Biol. 23, 60-62 Gold, J. M. and Gold, P. (1973) Cancer Res. 33, 2821-2824 Costanza, M. E., Pinn, V., Schwartz, R. S. and Nathanson, L. (1973) New Engl. J. Med. 289, 520-522 24 Hellstrom, 1., Hellstrola, K. E. and Shepard, T. H. (1970) Int. J. Cancer 6, 346-351 25 Hollinshead, A. C., Glew, D., Bunnag, B., Gold, P. and Herberman, R. B. (1970) Lancet, 1191-119¢

26 Hollinshead, A. C., McWrighL C. G., Alford, T. C., Gold, P. and Herberman, R. B. (1972) Science 177, 887-889 27 Lejtenyi, M. C., Freedman, S. O. and Gold, P. (1971) Cancer 28, 115-120 28 Thomson, D. M. P., Krupey, J., Freedman, S. O. and Gold, P. (1969) Proc. Natl. Acad. Sci. U.S. 64, 161-167 29 Goldenberg, D. M., Pavia, R. A., Hansen, H. J. and Vandevoorde, J. P. (1972) Nat. New Biol. 239, 189-190 30 Goldenberg, D. M. and Hansen, H. J. (1972) Science 175, 1117-1118 31 Dhar, P., Moore, T., Zamcheck, N. and Kupchik, H. (1972) J. Am. Med. Assoc. 221, 31-35 32 Holyoke, D., Reynoso, G. and Chu, T. M. (1972) Ann. Surg, 176, 559-564 33 Shuster, J., Silverman, M. and Gold, P. (1973) Cancer Res. 33, 65-68 34 Ekins, R. P. (1974) Br. Med. Bull. 30, 3-11 35 Hansen, H. J., Lance, K. P. and Krupey, J. (1971) Clin. Res. 19, 143 36 Egan, M. L., Lautenschleger, J. T., Coligan, J. E. and Todd, C. W. (1972) lmmunochemistry 9, 289-299 37 Tomita, J. T., Safford, J. W. and Hirata, A. A. (1974) Immunology 26, 291-298 38 Persijn, J. P., Korsten, C. B. and Renaud, J. (1974) Z. Klin. Chem. Kiln. Biochem. 12, 232 39 J. P. Persijn, personal communication 40 Bali, J. P., Fornajoux, J., Sebah, H., Balmes, J. L. and Marignan, R. (1973) Biol. Gastroenterol. (Paris) 6, 297-306 4l LoGerfo, P., Krupey, J. Hansen H. J. (1971) New Engl. J. Med. 285, 138-141 42 Go, V. L. W., Schutt, C. G., Moertel, C. G., Summerskill, W. H. J. and Butt, H. R. (1972) Gastroenterology 62, 754 43 MacSween, J. M., Warner, N. L., Bankhurst, A. D. and Mackay, I. R. (1972) Br. J. Cancer 26, 356-360 44 McPherson, T. A., Band, P. R., Grace, M., Hyde, H. A., Patwardhan, C. W. (1973) Int. J. Cancer 12, 42 54 45 Coller, J. A., Crichlow, R. W. and Yin, L. K. (1973) Cancer Res. 33, 1684-1688 46 Coligan, J. E., Egan, M. L. and Todd, C. W. (1972) Natl. Cancer Inst. Monogr. 35, 427-432 47 W. D. Terry, unpublished 48 Krupey, J., Wilson, T., Freedman, S. O. and Gold, P. (1972) Immunochemistry 9, 617-622 49 Rosai, J., Tillack, T. W. and Marchesi, V. T. (1972) Int. J. Cancer 10, 357-367 50 Kleinman, M. S., Harwell, L. and Turner, M. D. (1971) Gut 12, 1-10 51 Newman, E. S., Petras, S. E., Georgiadis, A. and Hansen, H. J. (1974) Cancer Res. 34, 2125-2130 52 Banjo, C., Shuster, J. and Gold, P. (1974) Cancer Res. 34, 2114-2121 53 Krupey, J., Gold, P. and Freedman, S. O. (1968) J. Exp. Med. 128, 387-398 54 Coligan, J. E., Lautenschleger, J. T., Egan, M. L. and Todd, C. W. (1972) lmmunochemistry 9, 377-386 55 Coligan, J. E., Henkart, P. A., Todd, C. W. and Terry, W. D. (1973) Immunochemistry 10, 591-599 56 Turner, M. D., Olivares, T. A., Harwell, L. and Kleinman, M. S. (1972)J. Immtmol. 108, 13281339 57 Rule, A. H. and Golesky-Reilly, C. (1973) lmmunol. Commun. 2, 213-226 58 Abeyounis, C. J. and Milgrom, F. (1972) lnt. Arch. Allergy Appl. Immunol. 43, 30 38 59 Terry, W. D., Henkart, P. A., Coligan, J. E. and Todd, C. W. (1974) Transplant. Rev. 20, 100-129 60 Marshall, R. D., and Neuberger, A. (1972) in Glycoproteins (Gottschalk, A., ed.), p. 351, Elsevier Publishing Co., Amsterdam

151 61 Banjo, C., Gold, P., Freedman, S. O. and Krupey, J. (1972) Nature 238, 183-185 62 Banjo, C., Gold, P., Gehrke, C. W., Freedman, S. O. and Krupey, J. (1974) Int. J. Cancer 13, 151-163 63 Coligan, J. E., Schnute, W. C., Egan, M. L. and Todd, C. W. (1974) U.S.A.E.C. Report Conf. 73-114l 64 C. Banjo, P. Gold, unpublished 65 Van Den Hamer, C. J. A., Morell, A. G., Scheinberg, 1. H., Hickman, J. and Ashwell, G. (1970) J. Biol. Chem. 245, 4397-4402 66 Thomas, P., Westwood, J. H. and Foster, A. B. (1974) Biochem. Soc. Trans. 2, 1248-1249 67 Terry, W. D., Henkart, P. A., Coligan, J. E. and Todd, C. W. (1972) J. Exp. Med. 136, 200-204 68 T. M. Chu, unpublished 69 Marshall, R. D. and Neuberger, A. (1964) Biochemistry 3, 1596-1600 70 Anderson, B. (1964) J. Biol. Chem. 239PC, 2716-2717 71 Banjo, C., Gold, P., Freedman, S. O. and Krupey, J. (1973) Fed. Proc. 32, 4444 72 Gold, J. M., Banjo, C., Freedman, S. O. and Gold, P. (1972) J. lmmunol. 111, 1872 1879 73 Marz, L., Bahl, 0. P. and Mohn, J. F. (1973) Biochem. Biophys. Res. Commun. 55, 717-723 74 Holborn, A. M., Mach, J. P, MacDonald, D. and Newlands, M. (1974) Immunology 26, 831-843 75 Hakkinen, 1. P. T. (1972) lmmunochemistry 9, 1115-1119 76 Burger, M. M. (1969) Proc. Natl. Acad. Sci. U.S. 62, 994-1001 77 Inbar, M. and Sachs, L. (1969) Nature 223, 710-712 78 Goldstein, 1. J., Hollerman, C. E. and Smith, E. E. (1965) Biochemistry 4, 87~883 79 Greenaway, P. J. and LeVine, D. (1973) Nature 241, 191-192 80 Kornfeld, S. and Kornfeld, R. (1969) Proc. Natl. Acad. Sci. U.S. 63, 1439-1446 81 Bird, G. W. G. (1959) Vox Sang. 4, 313-317 82 Hakomori, S. and Watanabe, K. (1973) FEBS Lett. 37, 317 83 Chu, T. M., Holyoke, E. D. and Murphy, G. P. (1974) Cancer Res. 34, 212-214 84 Burtin, P., Chavanel, G. and Hirsch-Marie, H. (1973) J. lmmunol. 111, 1926-1928 85 Mach, J. P. and Pusztaszeri, G. (1972) lmmunochemistry 9, 1031-1034 86 Kuo, T., Rosai, J. and Tillack, T. W. (1973) Int. J. Cancer 12, 532-542 87 Von Kleist, S., King, M. and Burtin, P. (1974) lmmunochemistry I1,249-253 88 Darcy, D. A., Turberville, C. and James, R. (1973) Br. J. Cancer 28, 147-160 89 Von Kleist, S. (1973) Ann. lmmunol. (Inst. Pasteur) 124C, 589-592 90 Turberville, C., Darcy, D. A., Laurence, D. J. R., Johns, E. W., Neville, A. M. (1973) lmmunochemistry 10, 841-843 91 Banjo, C. (1973) PhD Thesis, McGill University 92 Watkins, W. M. (1972) in Glycoproteins (Gottschalk, A. ed.), p. 830, Elsevier Publishing Co., Amsterdam 93 Khoo, S. K., Warner, N. L., Lie, J. T. and Mackay, 1. R. (1973) int. J. Cancer 11,681-687 94 Martin, F. and Martin, M. S. (1972) Int. J. Cancer 9, 641-647 95 Kupchik, H. Z. and Zamcheck, N. (1972) Gastroenterology 63, 95-101 96 Mach, J. P., Carrel, S., Merenda, C., Sordat, B. and Cerottini, J. C. (1974) Nature 248, 704-706 97 Khoo, S. K., Hunt, P. S. and Mackay, I. R. (1973) Gut 14, 545-548 98 Moore, T. L., Kupchik, H. Z., Marcon, N. and Zamcheck, N. (1971) Am. J. Dig. Dis. 16, 1-7 99 Hansen, H. J., Snyder, J. J., Miller, E., Vandevoorde, J. P., Neal Miller, O., Hines, L. R. and Burns, J. J. (1974) Human Pathol. 5, 139-147 100 Concannon, J. P., Dalbow, M. H. and Frich, J. C. (1973) Radiology 108, 191-193 101 Costanza, M. E., Das, S., Nathanson, L., Rule, A. and Schwartz, R. S. (1974) Cancer 33,583-590 102 Booth, S. N., King, J. P. G., Leonard, J. C. and Dykes, P. W. (1973) Gut 14, 794-799 103 Zamcheck, N., Moore, T. L., Dhar, P., Kupchik, H. Z. and Sorokin, J. J. (1972) Natl. Cancer Inst. Monogr. 35, 433-439 104 Reynoso, G., Chu, T. M. and Holyoke, D. (1972) J. Am. Med. Assoc. 220, 361-365 105 Moore, T. L., Dhar, P. and Zamcheck, N. (1972) Gastroenterology 63, 88-94 106 Band, P., Hyde, H., Feldstein, M., McPherson, T. A. and Patwardhan, V. (1973) Europ. J. Cancer 9, 597-602 107 Khoo, S. K. and Mackay, E. V. (1973) Aust. N.Z.J. Obstet. Gynaecol. 13, 107-113 108 Laurence, D. J. R., Stevens, U., Bettelheim, R., Darcy, D., Leese, C., Turberville, C., Alexander, P., Johns, E. W., Neville, A. M. (1972) Br. Med. J. iii, 605-609

152 109 Livingstone, A., Hampson, L. G., Shuster, J., Gold, P. and Hinchey, E. J. (1974) Archiv. Surg. 109, 259-264 110 Collaborative Study, Natl. Cancer Inst. Can., Am. Cancer Soc. (1972) Can. Med. Assoc. J. 107, 25-33 111 Gold, P., Wilson, T., Romero, R., Shuster, J. and Freedman, S. O. (1973) Dis. Colon Rect. 16, 358-365 112 Ona, F. V., Zamcheck, N., Dhar, P., Moore, T. L. and Kupchik, H. Z. (1973) Cancer 31,324-331 113 McCartney, W. H. and Hoffer, P. B. (1974) Radiology 110, 325-328 114 Moore, T. L., Kantrowitz, P. A. and Zamcheck, N. (1972) J. Amer. Med. Assoc. 222, 944-947 115 K.hoo, S. K. and Mackay, I. R. (1973) J. Clin. Pathol. 26, 470-475 116 Holyoke, E. D., Chu, T. M. and Murphy, G. P. (1974) Cancer, in the press 117 Steward, A. M., Nixon, D., Zamcheck, N. and Aisenberg, A. (1975) Cancer 33, 1246-1252 118 Vincent, R. G., Chu, T. M., Fergen, T. B. and Ostrander, M., unpublished data 119 Chu, T. M. and Nemoto, T. (1973) ,i. Natl. Cancer Inst. 51, 1119-1122 120 Cancannon, J. P., Dalbow, M. H., Liebler, G. A., Blake, K. E., Well, C. S. and Cooper, J. W. (1974) Cancer 34, 184-192 121 Khoo, S. K. (1974) Med. J. Austral. 1, 1025-1029 122 Zamcheck, N. (1974) Adv. Int. Med. 19, 413-433 123 Mach, J. P., Jaeger, P., Bertholet, M. M., Ruegsegger, C. H., Loosli, R. M. and Pettavel, J. (1974) Lancet ii, 535-540 124 Shuster, ,i., Livingstone, A., Banjo, C., Silver, H. K. B., Freedman, S. O. and Gold, P. (1974) Am. J. Clin. Pathol. 62, 243-257 125 Khoo, S. K. and Mackay, E. V. (1973) Aust. N. Z. ,i. Obstet. Gynaecol. 13, 1-17 126 Khoo, S. K. and Mackay, E. V. (1974) Cancer 34, 542 548 127 Vincent, R. G. and Chu, T. M. (1973) J. Thorac. Cardiovasc. Surg. 66, 320-328 128 Wechsler, M., LoGerfo, P., Feminella, ,i. and Lattimer, J. K. (1973) J. Urol. 109, 699-701 129 Ravry, M., Moertel, C. G., Schutt, A. J. and Go, V. L. W., unpublished data 130 Sorokin, J. J., Sugarbaker, P. H., Zamcheck, N., Pisick, M., Kupchik, H. Z. and Moore, F. D. (1974) ,i. Am. Med. Assoc. 228, 49-53 131 Mulcare, R. and LoGertb, P. (1972) J. Surg. Oncol. 4, 407-417 132 Skarin, A. T., Delwiche, R., Zamcheck, N., Lokich, J. J. and Frei, E. (1974) Cancer 33, 12391245 133 Laurence, D. ,i. R., Neville, A. M. (1972) Br. J. Cancer 26, 335-355 134 Hall, R. R., Laurence, D. J. R., Neville, A. M. and Wallace, M. D. (1973) Br. J. Urol. 45, 88-92 135 Delwiche, R., Zamcheck, N. and Marcon, N. (1973) Cancer 31,328-330 136 J. M. Gold and C. Banjo, unpublished 137 Kupchik, H. Z., Zamcheck, N. and Saravis, C. A. (1973) J. Natl. Cancer Inst. 51, 1741-1749 138 Smith. R. T. (1968) New Engl. T. Med. 278, 1268-1275 139 Smith, R. T. (1968) New Engl. J. Med. 278, 1326-1331 140 Lo Gerfo, P., Herter, F. P. and Bennett, S. J. (1972) Int. J. Cancer 9, 344-348 141 Von Kleist, S. and Burtin, P. (1966) Immunology 10, 507-515 142 Westwood, J. H., Bessell, E. M., Bukhari, M. A., Thomas, P. and Walker, .1. M. (1974) Immunochemistry I 1, 811-818 143 Westwood, ,i. H., Thomas, P. and Foster, A. B. (1974) Biochemical Society Transactions 2, 1250-1251