I in transformed and neoplastic cells and are responsible

Transcription

1 T-Antigen in Normal and Neoplastic Urothelium CATHERINE LIMAS, MD, AND PAUL LANGE, MDt The expression of the Thomsen-Friedenreich antigen (T-antigen) in normal and neoplastic urothelium was investigated using paraffin-processed and fresh frozen tissue sections obtained by biopsy from 56 patients. The T-antigen was detected through its binding to the peanut agglutinin (PNA) with two methods: a biotin-avidin-peroxidase system or a modified red cell adherence test. In vitro treatment of the tissue sections with neuraminidase induced PNA binding in the normal and neoplastic urothelium as well as the red blood cells and the vascular endothelium. Spontaneous PNA binding was absent in normal bladder epithelium, but was observed in 10% of the noninvasive and 65% of the invasive transitional cell carcinomas (TCC). The positive reactions were seen in the cytoplasm, cell surface, and mucin. These results were correlated with the light and electron microscopic findings and with the detectability of the A, B, H blood group antigens in the same tissues. Ultrastructurally, the PNA binding sites frequently corresponded to the Golgi apparatus and to secretory products which stained with Alcian blue. About 90% of TCCs with spontaneous PNA binding did not express the expected blood group antigen. The latter was also undetectable in 54% of TCCs lacking spontaneous PNA binding. It appears, therefore, that the expression of T-antigen occurs later in the evolution of aggressive TCCs, usually when they have advanced to invasive stages. The finding of spontaneous T-antigen unmasking correlates with the presence of invasion and a higher risk for metastatic involvement of regional lymph nodes. Cancer 58: , T IS WIDELY ACCEPTED that Cell Surface changes occur I in transformed and neoplastic cells and are responsible for the disruption of cell-to-cell interactions and altered responsiveness to hormonal and growth stimuli. Prominent among neoplastic cell surface abnormalities are the changes in antigenic composition of the cell membrane and/or cell coat which result either in the expression of neoantigens, and unmasking of cryptic and/or deletion of normal antigens. Such phenomena have been described in a variety of human neoplasms and attempts have been made to correlate their presence with the biological behavior of the tumor. In particular, the unmasking of cryptic determinants observed in a small number of human neoplasms, such as breast cancer, leukemia, and transitional carcinomas of the urinary bladder, deserves closer scrutiny. 1-5 We have chosen to examine the expression of Thomsen- Friedenreich antigen (T-antigen) in the normal and neoplastic urothelium because this antigen has been well characterized, can be detected in routinely preserved tis- From the Departments of *Pathology and turologic Surgery, Veterans Administration Medical Center and the University of Minnesota, Minneapolis, Minnesota. Supported by a grant (CA33239) from the National Cancer Institute, National Institutes of Health, Bethesda, MD. Address for reprints: Catherine Lima,, MD, Department of Pathology, VA Medical Center. 54th Street and 48th Avenue South, Minneapolis, MN Accepted for publication December 14, sue, and bears a close relationship to the degree of sialylation of glycoprotein and glycolipid molecules.6-8 The T-antigen becomes detectable on normal cells only after enzymatic removal of some sialic acid which can be accomplished in vitro through the use of neuraminidase. Furthermore, the demonstration of reduced sialic acid content in neoplastic urothelium makes it likely that some urothelial cancers spontaneously express the T-antigen.3,9 Detection of this antigen in tissues depends on the ability of a peanut agglutinin (PNA) to bind to the oligosaccharide D-galactosyl-P( 1-3)-N-acetyl-D-galactosamine carried by the T-antigen.8 In the current study, we compare the spontaneous and induced binding of PNA to normal and neoplastic urothelium and correlate the results with the detectability of tissue blood group antigens, histochemistry, and ultrastructure. Materials and Methods Fifty-six patients were included in the study in accordance with the following criteria: (1) tissue from one or more (average 2) biopsy specimens from the urinary tract was available for the required immunologic and histologic studies; (2) the diagnosis of transitional cell carcinoma (TCC) had been established in at least one of these specimens; (3) blood samples from all patients had been tested for A, B, 0 and Lewis a and b blood groups; and (4) one or more biopsy specimens from each patient had been tested for the presence of tissue A, B, or H antigens. 1236

2 No. 6 T-ANTIGEN IN UROTHELIUM LimUS UFZd Lunge 1237 The tissue samples were either frozen at -70" or preserved in paraffin. In 27 cases, both paraffin processed and frozen samples were examined. Tissue from 20 TCCs was also processed for electron microscopic study. In 39 patients (78 biopsy specimens), tissue binding of the Arachis hypogaea lectin (PNA) was performed using biotinylated PNA and an avidin-biotin-peroxidase complex (ABC). The biotinylated lectin and a kit containing Avidin DH and biotinylated horseradish peroxidase were purchased from Vector Laboratories Inc. (Burlingame, CA). In another 17 patients (34 biopsy specimens), the principle of the red cell adherence (RCA) test was applied to detect the T-antigen in tissues. The procedure using biotinylated PNA was camed out as follows: Sections were cut at 5 Fm, mounted on glass slides and washed for 5 minutes in buffer (0.01 M Tris- HC 1 ph 7.4,O. 1 5 M NaCl, M CaC12, MgC12). Two sections from each tissue block were incubated for 30 minutes at 22 C in the above buffer containing 0. I or 0.05 units of neuraminidase (Type V, Sigma) per milliliter. Another two sections from the same tissue block were incubated under the same conditions but without neuraminidase. Then, all four sections were washed in buffer and placed for 30 minutes in 0.3% H202 in methanol to inactivate endogenous peroxidases. After a 20-minute washing in buffer, the sections were covered with the appropriate dilution of PNA (25 pg/ml unless otherwise specified in Results) and incubated for 30 minutes at room temperature. Unbound lectin was then removed by washing the sections in buffer for 25 minutes. Following this, the sections were covered with Vectastain (ABC) for 60 minutes at room temperature. Excess Vectastain was washed away in buffer for 10 minutes and the sections were incubated with the diaminobenzidine-h202 solution (equal parts of 1 mg/ml diaminobenzidine in Tris-HC1 ph 7.2 and 0.03% H202). The development of the brown reaction product was observed under X 100 magnification. Further color development was stopped by immersing the slides in ice-cold water. The sections were counterstained with hematoxylin and cover-slipped using routine procedures. The reactions were classified as positive when more than 30% of the cells stained. When the positively stained cells did not exceed 30%, the reactions were classified as weak. The specificity of positive reactions was assessed in each case with the following controls: (1) sections incubated with non-biotinylated PNA which were negative; and (2) sections incubated with buffer, instead of PNA, which were negative. A modified RCA test for detecting T-antigen in tissues has been previously described by us in detail.3 Briefly, the sections were rinsed in buffer and then incubated with the extract of peanuts. After thorough washing in buffer, the sections were covered with a 2% suspension of T-ac- TABLE 1. Spontaneous PNA Binding in Normal and Neoplastic Urothelium by the Biotinylated PNA Method PNA Noninvasive Invasive reactions Normal Atypia CIS TCC TCC Positive 0 1 I 3 7 Weak positive Negative Total 30* t 17t * In three patients biopsy specimens were obtained from the bladder and the urethra, t Six patients had both noninvasive and invasive TCC. In three of these, both tumors were negative and in three, the invasive tumors only were positive. PNA: peanut agglutinin; CIS: carcinoma in situ; TCC: transitional cell carcinoma. tivated RBCs and incubated at room temperature. T-activated RBCs were prepared by treating normal donor RBCs with neuraminidase (0.1 unit/ml). The slides were then placed inverted on wooden applicator sticks in transparent plastic dishes containing normal saline so that the tissue sections were immersed in the fluid. The adherence of indicator RBCs was observed under XI00 magnification and the results were reported semiquantitatively as positive (50% or more of tissue reacting), weakly positive (30%-50% reacting) or negative to trace positive (less than 30% reacting). Duplicate sections were pretreated with neuraminidase as described above. The specificity of the RCA results were evaluated using (1) sections incubated with buffer instead of the peanut extract; and (2) sections incubated with unmodified, instead of T-activated, RBCs. In both procedures, the neuraminidase-treated sections served as positive controls, since the removal of sialic acid by this enzyme is expected to expose the D-galactosyl-/3 ( 1-3)-N-acetyl-D-galactosamine residues which bind the PNA.' The RBCs and the endothelium in all enzymatically treated tissues showed strong positive reactions, while they were invariably negative in the duplicate sections. The in vitro desialylation also induced PNA binding on normal urothelium, glands and mucous secretions. TABLE 2. Spontaneous PNA Binding in Normal and Neoplastic Urothelium by the T-Activated Red Blood Cell Adherence Method T-Activated Noninvasive Invasive RCA Normal TCC TCC Positive Weak positive Negative Total 10 II 6 PNA: peanut agglutinin; RCA: red cell adherence; TCC: transitional cell carcinoma.

3 1238 CANCER September Vol. 58 FIG. 1. Spontaneous PNA binding on the cell surface and in the cytoplasm of a papillary, noninvasive TCC, Grade 2. The sites of PNA binding are visualized by a brown reaction product (N = nucleus) (biotinylated PNA-Vectastain-hematoxylin, X250). Detection of tissue blood group A, B and H antigens was accomplished using the RCA test as previously reported." Selected tissue blocks from 20 TCCs were studied by electron microscopy. Sections were cut from the plastic (Poly/Bed 8 12) embedded blocks at 80 nm, stained with uranyl acetate and lead citrate and examined with a Zeiss 1 OC electron microscope. Hematoxylin-eosin (H & E)- stained sections were evaluated for morphologic characteristics such as grade and invasive growth. The presence of mucin, whenever relevant, was confirmed using the alcian blue-pas reaction at ph 2.5. Results Of the 56 patients studied, 32 had papillary noninvasive TCCs, 17 invasive TCCs, 6 both papillary noninvasive and invasive TCCs, and 1 flat carcinoma in situ (CIS) only without discrete tumor mass. In addition to the tumor mass, CIS was found in four patients and severe to TABLE 3. Localization of the Spontaneous Biotinylated PNA Binding Noninvasive PNA reactions positive in Normal Invasive TCC Cytoplasm Cell surface 0 I 0 Cytoplasm and cell surface FIG. 2. Spontaneous PNA binding in an invasive TCC, Grade 3. Many Total of the neoplastic cells stain dark brown. Notice that the vascular endo- PNA: peanut agglutinin; TCC: transitional cell carcinoma. thelium and the RBC in the vessel (arrow) show no spontaneous PNA binding (biotinylated PNA-Vectastain-hematoxylin, X40).

4 No. 6 T-ANTIGEN IN UROTHELIUM * Limus und Lunge 1239 moderate atypia in seven. Morphologically unremarkable bladder mucosa was studied in 37 of these patients, while normal urethral mucosa was also biopsied in three cases. Table 1 summarizes the results obtained with biotinylated- PNA and Table 2 the results of the RCA test for the T- antigen without in vitro desialylation. The morphologically normal bladder mucosae, von Brunn nests and their mucus were invariably negative for spontaneous PNA lectin binding with either method. Similarly, the vascular endothelium and the red blood cells within the tissues were nonreacting in all biopsies. Spontaneous positive reactions were observed in the normal appearing urothelium of one and in the periurethral glands of two urethral biopsy specimens. Three of the 27 noninvasive papillary TCCs evaluated with the biotinylated PNA method and one of the 11 similar tumors studied with the RCA test gave positive spontaneous reactions for PNA binding. These reactions were patchy in various areas of the tumor and could be localized in the cytoplasm (one case) or in both cytoplasm and cell surface (two cases) as illustrated in Figure 1. As shown in Table 3, there was a high frequency of PNA positive staining among invasive TCCs. Thus, 1 1 of the 17 invasive tumors evaluated with the biotinylated- FIG. 4. Spontaneous intracytoplasmic PNA binding in an invasive TCC. Notice the sharply demarcated intense staining in the paranuclear regions (biotinylated PNA-Vectastain-hematoxylin, X250). FIG. 3. Spontaneous PNA binding on the surface of some neoplastic cells in CIS that extends into von Brunn nests (biotinylated PNA-Vectastain-hematoxylin, X 100). PNA and 4 of the 6 tested with the RCA gave positive spontaneous reactions (Fig. 2). Overall spontaneous PNA binding was observed in 10% of the noninvasive and 65% of the invasive TCCs. Among cases with coexisting invasive and noninvasive TCCs, only the former was positive in three while the latter was negative in all six. One of the five carcinomas in sitzi and one of the seven atypical mucosas showed positive staining (Fig. 3). The localization of the reactions in the 1 1 positive invasive TCCs was variable, usually being both in the cytoplasm and on the cell surface (six cases), less often in the cytoplasm only (four cases) and rarely exclusively on the cell surface (one case). These results are summarized in Table 3 and illustrated in Figures 4 and 5. In six TCCs the well defined intracellular PNA-staining was shown to correlate with positive alcian blue staining for mucin. In vitro incubation of the tissue sections with neuraminidase resulted in PNA binding by the endothelium, red blood cells and morphologically unremarkable urothelium (Figs. 6A and 6B). Of the 30 normal mucosas, 24 became positive and 6 weakly positive. All cases of CIS and atypia also became positive following treatment with the enzyme. Similarly, all TCCs gave positive or weakly positive reactions following desialylation. Spon-

5 1240 CANCER September Vol. 58 FIG. 5. Spontaneous PNA binding on the cell surface of an invasive TCC (biotinylated PNA-Vectastain-hematoxylin, X250). taneously reacting TCCs retained their ability to bind the PNA even after neuraminidase treatment (Table 4). The localization of positive staining following enzymatic removal of sialic acid was usually both intracellular and on the cell surface (21 normal mucosas and 33 TCCs), sometimes intracellular only (9 normal and 6 TCCs) and least often on the cell surface only (5 TCCs) (Table 5). In general, after desialylation, the positive staining became more widespread and intense. In particular, enzyme treatment resulted in intense cell surface staining in 34 TCCs, only 5 of which expressed spontaneous PNA binding on cell surfaces (Figs. 7A and 7B). Also, the neuraminidase intensified the cell surface reactivity in the eight TCCs which had spontaneous reactions in both cytoplasm and cell surface. Very consistently, enzymatic desialylation induced strong positive reactions in extracellular as well as intracellular mucous secretions in both normal and neoplastic tissues. When fresh-frozen tissues were examined, the results correlated with those obtained with paraffin processed material. However, it was noted that the reactions in fresh tissues were less clearly localized, in particular following neuroaminidase treatment, and did not permit good differentiation between the various staining patterns which were evident in the processed sections. The ultrastructural studies confirmed the presence of FIGS. 6A AND 6B. (A) Normal bladder mucosa incubated with biotinylated PNA and Vectastain and counterstained with hematoxylin. There is no positive spontaneous reaction in either the epithelium or vessels. (B) The same tissue when pretreated with neuraminidase showed intense reactions on the epithelium, vascular endothelium and RBC (X40).

6 ~~ No. 6 T-ANTIGEN IN UROTHELIUM - Limas and Lange 1241 TABLE 4. Neuraminidase Induced PNA Binding in Normal and Neoplastic Urothelium by the Biotinylated PNA Method PNA Noninvasive Invasive reactions Normal Atypia CIS TCC TCC Positive Weak positive Negative PNA: peanut agglutinin; CIS: carcinoma in situ; TCC transitional cell carcinoma. TABLE 5. Localization of Biotinylated PNA Binding After In Vim Desialylation of Normal and Neoplastic Urothelium Invasive Noninvasive PNA reactions positive in Normal TCC TCC Cytoplasm Cell surface Cytoplasm and cell surface Total PNA: peanut agglutinin; TCC: transitional cell carcinoma. secretory activity and glandular differentiation which corresponded to the most intense staining for PNA and alcian blue. As shown in Figures 8A-8C, the well defined intracytoplasmic regions which stained spontaneously for PNA could be correlated with prominent Golgi apparatus and intracellular lumina. In addition, microacinar structures were identified in TCCs which had very intense localized reactions following in vitro removal of sialic acid (Figs. 9A-9C). TCCs with spontaneous PNA binding were usually (87%) nonreactive for the expected BG antigen. However, the expected BG antigen was also undetectable in about 54% of tumors lacking spontaneous PNA binding (Table 6). All invasive TCCs were classified as Grade 3 or higher. Of the noninvasive TCCs one was Grade 1, 15 Grade 2 and 11 Grade 3. Two Grade 3 (18%) and one Grade 2 (6.7%) noninvasive TCCs gave spontaneous PNA staining. Thus, spontaneous PNA binding appeared to correlate best with invasiveness. The 17 patients with invasive TCCs were treated with cystectomy or radiation followed by cystectomy. Five of these patients died of the bladder cancer and of these four had spontaneously PNA reacting tumors. Metastases in the pelvic lymph nodes were found in 60% of the PNApositive and none of the PNA-negative invasive TCCs. The three patients with spontaneously PNA-positive noninvasive TCCs are alive without evidence of progression to higher stages one to 5 years later. FIGS. 7A AND 7B. (A) TCC Grade 2-3 shows no spontaneous PNA binding. (B) The Same TCC shows mostly cell surface PNA binding after treatment with neuraminidase. Notice that the vascular endothelium and the RBC also became positive (X 100).

7 1242 CANCER September Vol. 58 FIGS. 8A-8C. (A) Spontaneous PNA binding in the cytoplasm and in extracellular much (arrows) of an invasive TCC (X250). (B) Ultrastructure of the same tumor shows prominent Go18 complexes with good correspondence to the localized paranuclear PNA binding (uranyl acetate-lead citrate, X3500). (C) Luminal structures containing secretory products which correspond well to the pools of PNA positive material which also stained with alcian blue (Stain as in B, X3500). Discussion Many antigenic determinants present at the cell surface are chemically glycoproteins and glycolipids and their specificity is determined by terminal oligosaccharides. A large variety of such determinants can be readily detected on unmodified normal cells by the use of specific antibodies and/or lectins. Additional antigenic groups may become available for antibody and/or lectin binding following enzymatic modification of the cell surface mo- lecular structure. Such normally concealed determinants are usually referred to as cryptic or masked antigens. In vivo, spontaneous unmasking of some cryptic antigens has been observed in pathologic states and has attracted some attention in the field of cancer research. -4 Particularly, the role of sialylation of cell surface macromolecules in modifying the accessibility of antigenic determinants is under investigation because of the multifaceted function of the sialic acids, their high concentration at the surface of mammalian cells, and their abnormal me-

8 No. 6 T-ANTIGEN IN UROTHELIUM - Limas and Lange 1243 FIGS. 9A-9C. (A) PNA binding after neuraminidase treatment of a papillary noninvasive TCC. There is positive staining on the cell surface, in paranuclear locations and in rounded pools (arrows) (B) Ultrastructure of the same tumor shows prominent Golgi areas probably corresponding to the paranuclear PNA binding sites (Uranyl acetatelead citrate, X3500). (C) A microacinar structure with a miniature lumen, which may correspond to the intensely PNA positive pools of mucin that were stained with alcian blue (Stain as in B, X2600). tabolism in neoplastic and transformed cell^.'^-'^ A wellknown cryptic antigen which becomes apparent in a number of normal cells when subjected to desialylation, is the T-antigen. This was originally observed on RBCs of bacterially contaminated blood which were agglutinated by all normal human sera and were later found to bind a lectin found in the extract of peanuts (arachis hypo- gaea).6,15 It was also shown that RBCs, as well as several other cell types, acquired the same properties when treated in vitro with neuraminidase, a sialic acid-removing enzyme. Subsequently, spontaneous unmasking of the T- antigen was reported in cancer cells from breast and unnary bladder.',3,4 These observations, together with some biochemical evidence of defective sialylation of the cell

9 1244 CANCER September Vol. 58 TABLE 6. Correlation Between BG Antigen and Spontaneous T-Antigen Reactivity in Transitional Cell Carcinomas Spontaneous PNA reactions* Reactivity for the expected BG antigen? Positive Negative Positive 2 14 Negative Total * Tested by the biotinylated PNA lectin method on paraffin processed tissues. Tested by the RCA method on paraffin processed tissues. PNA: peanut agglutinin; BG: blood group. surface in malignant transformation, led to the speculation that the detectability of T-antigen reflects deficient sialic acid incorporation onto the macromolecules which carry the PNA binding sites. Because disturbances in sialylation would result in altered cell surface charge and cell-to-cell interaction, as reported for some transformed cells, it is plausible that spontaneous unmasking of the T-antigen may serve as an indicator of malignant potential. We have previously examined this hypothesis by studying the expression of the T-antigen on normal and neoplastic unothelium in conjuction with alterations in the detectability of the tissue blood group (BG) antigens A, B and H.3 We had concluded that the neoplastic cells were deficient in sialic acid and that the T-antigen was frequently accessible in urothelial tumors which had become invasive. Other investigators have also suggested that the expression of this cryptic antigen in TCCs of the urinary bladder may be used as a marker of poor progn~sis.~ The current communication is a detailed analysis of the PNA-binding properties of a large number of normal bladder mucosas and TCCs. The binding of PNA is correlated with conventional morphology, histochemistry, and ultrastructure and a hypothesis is developed which offers an explanation of our findings. The results clearly demonstrate that normal as well as neoplastic urothelium possess cryptic PNA binding sites which are exposed by the action of neuraminidase, i.e., when some sialic acid is removed. Furthermore, our results show that spontaneous unmasking of such sites is a phenomenon frequently occurring in morphologically malignant cells of urothelial origin which have developed the ability to invade the adjacent tissues. This phenomenon is occasionally observed in noninvasive papillary TCCs, but must be extremely exceptional in nonneoplastic bladder mucosa. Of the 40 patients who had nonneoplastic urothelium examined (total of 65 biopsy specimens) only one showed positive PNA staining without in vitro desialylation. Interestingly, this biopsy specimen was from the prostatic urethra, which suggests that topographic differences may exist in the expression of antigens of morphologically similar cells. The spontaneous PNA reactions are often seen on both the cell surface and in the cytoplasm (about 57% of cases) and variable patterns are common within the same tumor. Exclusively, cell surface staining is seldom encountered as a spontaneous phenomenon but, following desialylation, the cell surface staining is intensified so that the neuraminidase treated sections often show a predominance of this pattern. The intracellular staining is often confined to a distinct paranuclear area which corresponds well to the Golgi apparatus. In addition, intense, well-defined intracellular reactions can be shown to correlate with the alcian blue staining for acid mucopolysaccharrides and with ultrastructural evidence of intracellular secretory products. These patterns of intracellular PNA binding were also seen in most invasive and noninvasive TCCs following incubation with neuraminidase. Moreover, in vitro induced PNA binding was very intense in localized extracellular areas which also reacted for acid mucopolysaccharides and corresponded to microacinar formations at the ultrastructural level. The specificity of these reactions for the lectin was confirmed by the absence of staining in the negative controls. Whether these PNA binding sites are identical to the T-antigen cannot be proven by either the avidin-biotin or the peroxidase-antiperoxidase method. The best evidence that the reacting components are closely related, if not identical, to the T-antigen is provided by the RCA test, since the indicator RBCs can adhere to the tissue only through the exposed T-antigen on their surface. The major disadvantage of the RCA test is its poor resolution between surface and intracellular reactions. It has been reported that some malignant TCCs cannot bind the PNA even after treatment with ne~raminidase,~ which is at variance with our results. Since these authors have used different methodology and different conditions of desialylation, the results of the two studies are not strictly comparable. At present, we offer the following hypothesis which may explain our findings. Macromolecules that carry the PNA binding sites are transported to the cell surface after passing through the Golgi apparatus, where they may reach the concentration necessary for visualization with our methodology. In addition, similar binding sites are found on molecules which can be secreted and are, therefore, present within secretory vacuoles and extracellular secretions. If the sialylation of the carrier molecules is completed to the degree of masking the PNAbinding sites, the reactions have to be induced by enzymatic removal of some sialic acid. Therefore, in normal bladder epithelium, positive reactions will appear in any of the above locations only after neuraminidase treatment. In case of abnormal sialylation, spontaneous positive reactions would appear in the same locations if the transport of the molecules were not affected by the addition of sialic acid. Our findings suggest that sialylated molecules are preferentially exteriorated, so that the excreted material

10 No. 6 T-ANTIGEN IN UROTHELIUM * Limas and Lange 1245 and the cell surface may retain the masking of the PNAbinding sites, whereas spontaneous unmasking is noted in the Golgi or in intracellular secretions of neoplastic cells. These findings are consistent with biochemical studies which have localized sialylating and glycosylating activities in the Golgi apparatus. 6,17 Since neuraminidase intensified the reactions in all cases, we assumed that some degrees of sialylation does take place even in the invasive tumor cells. TCCs with spontaneous PNA reactions are frequently negative for the expected BG antigen, but the two phenomena are probably not directly related because about 50% of the BG antigen-negative TCCs do not express PNA binding. The loss of normal antigenic determinants is distinct from the unmasking of cryptic sites, but the two processes coincide in the more aggressive tumors the latter probably occumng later in their biologic course. These results also suggest that the spontaneous unmasking of T-antigen in invasive TCCs correlates with a greater metastatic potential. However, this suggestion needs to be evaluated by examining more cases in a prospective study which we have currently undertaken. REFERENCES 1. Springer GF, Desai PR, Banatwala 1. Blood Group MN antigens and precursors in normal and malignant human breast glandular tissue. J Natl Cancer Inst 1975; 54: Reisner Y, Binaminov M, Rosenthal E, Sharon N, Ramot B. Interaction of peanut agglutinin with normal human lymphocytes and with leukemic cells. Proc Natl Acad Sci US : 76: I. 3. Limas C, Lange P. Altered reactivity for A,B,H antigen in transitional cell carcinomas of the urinary bladder: A study of the mechanisms involved. Cancer 1980: 46: Coon JS, Weinstein RS, Summers YL. Blood group precursor T- antigen expression in human urinary bladder carcinoma. Am J Clin Pathol 1982: Newman RA, Klein PJ, Rudland PS. Binding of peanut lectin to breast epithelium, human carcinomas and a cultured rat mammary stem cell: Use of the lectin as a marker of mammary differentiation. J Natl Cancer Inst 1979; 63: Bird GW. Anti-T in peanuts. Vox Sang 1964: 9: Lotan R, Skutelsky E, Danon D, Sharon N. The purification, composition and specificity of the anti-t lectin from peanuts (Arachis hypogaea). J Biol Chem 1975; Lotan R, Sharon N. Peanut (Arachis hypogaea) Agglutinin; In: Ginsburg V ed. Methods in Enzymology, vol. 50. New York: Academic Press 1978; Dermer GB, Kern WH. Changes in the affinity of phosphotungstic acid and positively charged colloidal particles for the surfaces of malignant human transitional epithelium of the urinary bladder. Cancer Res 1974; 34: Limas C, Lange P, Fraley E et al. A,B,H antigens in transitional cell tumors of the urinary bladder: Correlation with the clinical course. Cancer 1979; 44: I 1. Hughes RC. In: Hughes RC, ed. Glycoproteins. New York Chapman and Hill, 1983: Lloyd CW. Sialic acid and the social behavior of cells. Bid Rev 1975; 50: Grimes WJ. Sialic acid transfemes and sialic acid levels in normal and transformed cells. Biochemistry 1970; 9: Jeanloz RW, Codington JF. Biological roles of sialic acid. In: Rosenberg A, Schengrund CL, eds. New York: Plenum, 1976; 20 I. 15. Friedenreich V. The Thomsen Haemagglutination Phenomenon. Copenhagen: Levin and Munksgaard, Schauer, R. Biosynthesis of sialic acids. In: Ginsburg V, ed. Methods in Enzymology, vol. 50, part C. New York: Academic Press, Berger EG, Mandel T, Schilt V. Immunochemical localization of galactosyl transferase in human fibroblasts and HeLa cells. J Histochem Cytochem 1981: 29: