ABO(H) antigens and beta-2 microglobulin in transitional cell carcinoma. Predictors of response to intravesical bacillus calmette-guerin

ABO(H) Antigens and Beta-2 Microglobulin in Transitional Cell Carcinoma Predictors of Response to Intravesical Bacillus Calmette-Guerin Holt Sanders, MD, Peter McCue, MD, and Sam D. Graham Jr, MD The response of patients with superficial transitional cell carcinoma of the bladder (STCB) to intravesical chemotherapy is variable; some patients enjoy a long period without recurrence, whereas others have recurrence of tumor within 2 years of removal of the primary lesion. Previously, others have demonstrated that the loss of normal cell surface antigens, such as ABO(H) blood group antigens or beta-2 microglobulin (B2M) has been correlated with more aggressive behavior by tumor. In this study, using immunohistochemical techniques, the authors evaluated the initial pretreatment biopsy specimen of bladder tumors for the presence of ABO(H) antigens and B2M. Data from this sample patient population, all with biopsy-proven STCB, indicate that expression of these two markers is predictive of a therapeutic response to prophylactic intravesical bacillus Calmette-Guerin (BCG) (Tice strain) after resection, and that expression of the two markers is of greater predictive value than expression of either antigen alone. Cancer 67:3024-3028,1991. HE NATURAL HISTORY of superficial transitional cell T carcinoma of the bladder is extremely variable. Of all cases that occur annually in this country, approxi- mately 50% will recur within 2 years after fulguration, and 10% of these patients will have a tumor recurrence of a higher stage.' Intravesical chemotherapy of superficial transitional cell carcinoma of the bladder with such agents as thiotepa and bacillus Calmette-Guerin (BCG) has been shown in the last 20 years to be extremely effective in reducing the recurrence rate. Veneema et aL2 showed that thiotepa was effective in the treatment of bladder cancer, and for many years it was the most popular intravesical agent.2 In 1977, the National Bladder Cancer Collaborative Group3 reported that 69% of patients given a regimen of thiotepa were tumor free at 2 years compared with 26% of the control group. In 1976, Morales et aL4 published a report of nine From the Departments of Surgery (Urology) and Pathology. Emory University School of Medicine, Atlanta. Georgia. The authors thank Dr. Michael Kutner for statistical consultation and Ms. Cynthia Roseberry for her help in the preparation of this manuscript. Address for reprints: Sam D. Graham, Jr. MD, Department of Surgery, Division of Urology, Emory Clinic. 1365 Clifton Road. N.E.. Atlanta, GA 30322. Accepted for publication November 5. 1990. patients with superficial bladder cancer, all of whom re- sponded to intravesical BCG therapy. Many reports since then have supported the finding that BCG is extremely efficacious in the treatment of this disease. A study by Brosman5 in 1981 found that in the 27 patients randomized to BCG therapy there were no recurrences in a 24-month period. Ofthe 19 patients who received thiotepa, there were nine recurrences. Schell- hammer and associates' published a report of 28 patients with recurrent superficial bladder cancer who received BCG, of whom 20 (7 1%) showed no histologic or cytologic evidence of tumor in follow-up examinations. A study comparing resection and/or fulguration of superficial bladder cancer with resection and/or fulguration plus BCG instillation showed a response rate of 62% after 3 months in the BCG group compared with a response rate of 39% after 3 months in the control group.7 Although the therapeutic effects of BCG have been im- pressive, there is still the question of why, in a group of patients who have the same disease morphologic type, some patients do not respond. Lamm et al.' correlated response to BCG therapy with conversion of purified pro- tein derivative (PPD) skin testing from negative to posi- tive. A granulomatous reaction is seen in a proportion of cases treated with BCG, as well as in some bladders after 3024 No. 12 BLADDER CELL SURFACE MARKERS - Sanders et al. 3025 biopsy or other manipulations. These reactions tend to be transitory and focal in nature. One study reported that 77% of bladders with granulomatous reaction to the in- travesical therapy had a complete response, whereas only 32% of bladders lacking a granulomatous reaction re- sponded to therapy.' It was postulated that some patients responded because they were able to mount an adequate immune response. Another possibility is that those who fail to respond have a more aggressive lesion, one that cannot be easily differentiated from a less aggressive lesion on the basis of histologic study. As these morphologic changes are merely later expressions of prior molecular events, it would appear that examination of molecular changes in these tumors could distinguish the biologic differences between two morphologically similar entities, and would therefore be more sensitive in predicting aggressive behavior than his- tologic study alone. Many studies in the last few years have shown that the loss of ABO(H) blood group antigens from the surface of bladder cells correlates with the development of malig- nancy and is, in some instances, predictive of aggressive behavior ofthe t ~ m o r . ' ~ - ' ~ The loss of antigens may even precede the morphologic changes of malignancy, and his- tologically benign but antigenically abnormal urothelium may indicate a premalignant l e ~ i o n . ' ~ . ' ~ Other blood group-related antigens, for example, Lewis antigens and the T antigen, and the unmasking of the T antigen (cryptic T) have also been correlated with the development of neopla~ia.'~.'' These antigens all share a common biosyn- thetic pathway with the ABO(H) antigens, and their loss is therefore not completely independent of ABO(H) an- tigen loss. Walton et al." have shown that B2M, a molecule which accompanies the class I major histocompatibility complex (MHC) antigens on the surface of all mammalian cells, also disappears with progressive malignant transforma- tion. The biosynthetic pathway of B2M is entirely separate from that of the blood group-related antigens, and its expression is therefore independent of ABO(H) antigens. Here we report a study involving bladder biopsy spec- imens of patients with superficial transitional cell carci- noma of the bladder before they received intravesical che- motherapy to determine whether the expression of two separate tumor markers, ABO(H) antigens and B2M, is predictive of a therapeutic response. Data from our sample indicate that loss of these cellular antigens is of prognostic significance, and that expression of the two markers is of greater predictive value than the expression of either an- tigen alone. Materials and Methods Fifteen patients treated with BCG for superficial tran- sitional cell carcinoma between 1984 and 1985 at Emory University (Atlanta, GA) had tissue from tumor resection before BCG available. Patients on therapy for 24 months without a recurrence were considered responders. Patients with a recurrence within 24 months of the beginning of therapy were considered nonresponders. Formalin-fixed, paraffin-embedded tissues representing the bladder tumors before the beginning of intravesical chemotherapy were obtained for immunoperoxidase studies to demonstrate the presence and distribution of the B2M and the ABO(H) blood group antigens.21,2' Histologic Criteria Serial unstained sections were cut from the formalin- fixed, paraffin-embedded tissues. Standard hematoxylin and eosin stains were prepared on consecutive slides and evaluated for urothelial lesions. Tumors were pathologi- cally graded and staged according to the TNM classifi- cation. These findings were compared with the original evaluations. There was concordance on the histopatho- logic evaluations. Basic Immunoperoxidase Technique The technique of immunoperoxidase staining of tissue by the labeled secondary antibody has been previously de~cribed.'~ Briefly, 4-pm sections were cut and deparaf- finized with xylene and graded ethanol baths. The tissue was rehydrated in 0.10 mol/l phosphate-buffered saline, pH 7.5, and endogenous peroxidase activity was inhibited by incubation in a mixture of 200 ml methanol and 50 ml of 1% hydrogen peroxide for 20 minutes. The speci- mens were then treated with diluted normal goat serum for 20 minutes at room temperature to block nonspecific staining. Beta-2 Microglobulin The technique for staining paraffin-fixed sections for B2M has been previously described.20 Briefly, the tissues were incubated for 40 minutes with primary rabbit anti- B2M serum diluted 1:60 with 1% bovine serum albumen in phosphate-buffered solution. They were incubated for 30 minutes in biotinylated goat anti-rabbit serum, IgG fraction, diluted 1:200, followed by incubation for 30 minutes in avidin-biotin-peroxidase complex solution, diluted 1:50. All incubations were performed at room temperature. A freshly prepared solution of 0.65% di- aminobenzidine mixed 1 : 1 with 1 % hydrogen peroxide was used for the chromogen reaction. The tissues were then counterstained with hematoxylin and mounted with Permount. Original titration of the antibodies was performed on normal biadder sections from random biopsy cases and cystectomy specimens. Each run included a normal con- 3026 CANCER Jirne 15 199 1 Vol. 67 trol which consisted of deleting the primary antibody on parallel sections. ABO(H) Blood Groiip Antigens The technique for ABO(H) staining followed the pro- tocol of Coon and Weinstein.” Those tissues stained for the blood group antigens were also deparaffinized, rehy- drated, and treated with methanol-hydrogen peroxide so- lution to block background staining. The tissue samples were then incubated for 16 hours with primary mouse antibody to blood group antigens A, B, and H, diluted 1: I0 with 1% bovine serum albumin in phosphate-buffered solution. Incubations were performed in a humidified chamber at room temperature. The samples were incu- bated for 30 minutes in biotinylated goat anti-mouse serum, IgM fraction. The procedure for activation of the immunoperoxidase was identical to that described above in the staining for B2M. Evaluation of Immiinohistochemical Staining As most tissue sections had lesions of differing severity as well as areas of normal urothelium, the lesion of greatest severity was used in scoring. The staining characteristics of adjacent areas were also recorded and were helpful in distinguishing our one nonsecretor. For example, a tissue section might have an area of normal urothelium which stained very intensely and an area of neoplasia which had no staining whatsoever in any layers of the urothelium. Focally, the urothelium showed complete loss of both ABO(H) blood group antigens and B2M but represented a secretor. If all urothelium, diseased as well as normal, failed to stain for ABO(H) antigens, the patient was con- sidered a nonsecretor. Staining was evaluated for intensity, pattern (basilar or apical), and extent per lesion field. In tabulating the data, if a lesion had at least partial staining, it was scored as positive. TABLE I Patient no. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Grade Stage ABO(H) Beta-2 Response Secretor - 1 Ta I1 Ta 11 Ta I1 Ta + 11 Ta + 11 Ta + 11 Ta + I1 Ta + 111 Ta I11 Ta + 111 Tis 111 Tis I1 T1 11 TI 111 T1 - ~ - - - - - - + + + + + + + + - - - + + + - + + t + + + + + t + + + + + ~ TABLE 2. Correlation of Stain to Response ABO(H) Beta-2 Resp NResp Total Resp NResp Total 3 Positive stain 4 6 5 6 1 1 Negative stain I 8 9 0 4 4 Total 5 10 15 5 10 15 Resp: 2 yr without a recurrence; NResp: had a recurrence of tumor within 2 yr. Statistical Anal-vsis 2 X 2 tables. Fisher’s exact test was used to determine P values for Results The data for all 15 patients are shown in Table 1. Of nine patients with a negative stain for ABO(H), that is, with no stain at all on the lesion, one (1 1%) responded to intravesical chemotherapy. There were six patients whose lesions stained positively for ABO(H), and four (67%) of these patients responded to chemotherapy (Table 2, P = 0.047). Of the four patients whose staining showed the absence of B2M, none responded to chemotherapy ( P = 0.0625). B2M was present on the lesions of 1 1 patients, five (45%) of whom responded to chemotherapy. A combined evaluation of the two markers has a greater prognostic significance than either one alone. There were five tumors that stained positively for both ABO(H) an- tigens and B2M and four (80%) of these responded to chemotherapy. Of the ten tumors that stained positively for only one or neither ofthe antigens, nine (90%) recurred within 2 years (Table 3, P = 0.017). Discussion The data in this study indicate that the loss of both ABO(H) antigens and B2M on bladder tumor cells is strongly predictive of a poor response to intravesical che- motherapy with BCG, whereas the presence of both an- tigens is predictive of a good response. Although the loss TABLE 3. Use of Two Antigens to Predict Response 2 Ag Pos* 1 or 0 Ag Post Responders 4 Nonresponders I Total 5 1 9 10 Ag: antigens; Pos: positive. P = 0.017. * Both beta-2 microglobulin and ABO(H) antigens were present on the cell surface. t Only beta-2 microglobulin were present, or only ABO(H) antigens were present, or neither antigen was present on the cell surface. No. I2 BLADDER CELL SURFACE MARKERS Sanders ef al. 3027 of ABO(H) antigens is the more sensitive predictor of re- currence, the loss of B2M is the more specific predictor. This predictive accuracy is independent of tumor grade. The staining characteristics of the two markers is even more predictive when evaluating lesions restricted to Stage Ta and Tis (Table 4). Several previous investigators have found a significant correlation between the expression of ABO(H) blood group antigens and recurrence of tumor. In a study by Bergman and Javadpour," seven of nine patients negative for ABO(H) had a recurrence within 5 years, whereas in the same study, four of six patients positive for ABO(H) had no recurrence in 5 years. In a comparing of ABO(H) expression to response to intravesical therapy in one study, 13 of 1 8 (72%) ABO(H)-negative tumors recurred after therapy; however, the more aggressive ABO(H)-negative tumors recurred less frequently than the ABO(H)-positive tumors.8 Lange et a1.I' found that of 16 patients who stained positively for ABO(H) antigens, two (13%) later developed invasive disease; of 21 patients who failed to stain for ABO(H) antigens, 16 (76%) eventually developed invasive disease. The predictive value of ABO(H) antigen staining in the above-mentioned study was independent of cytologic grade. D'Elia and associate~'~ reported 2 1 of 25 tumors (84%) with ABO(H) antigens remained at Stage TO for a mean follow-up of 13 years. Eleven of 15 tumors (73%) that had lost the expression of ABO(H) antigens progressed to stage T2 or greater in a mean of 4 years. All Grade 1 and Grade 2 lesions in this study had at least apical staining for B2M. Positive B2M staining was not predictive of successful therapy in the lower grade lesions (unless accompanied by positive staining for ABO(H) antigens), but loss of B2M correlated with lack of response. Of the four tumors that were negative for B2M, all recurred within 2 years. Only one of the B2M- negative tumors stained positively for ABO(H) antigens. This finding suggests that urothelium undergoing malig- nant transformation may lose ABO(H) blood group an- tigens before it loses B2M. There is good evidence that BCG acts against tumors by stimulating T-lymphocytes and macrophages.' Perhaps the B2M component of the MHC-1 antigen complex is required for immune recog- TABLE 4. Stage Ta-Tis 2 Ag Pos* I of 0 Ag Post Responders 4 Nonresponders 1 Total 5 0 7 I Ag: antigens; Pos: positive. P = 0.01. * Both beta-2 microglobulin and ABO(H) antigens were present on the cell surface. t Only beta-2 microglobulin were present, or only ABO(H) antigens were present, or neither antigen was present on the cell surface. nition of the tumor, with the loss of B2M and the MHC- 1 antigen complex resulting in an inadequately focused immune response. This mechanism might explain the high specificity of B2M loss as a predictor of recurrence after BCG therapy.24 In 12 of 15 (80%) biopsies stained in this study, there were areas of normal urothelium with normal staining for antigens adjacent to the malignant and dysplastic tis- sues. This observation is at variance with the finding by Stein25 that between 56% and 75% of histologically normal urothelium has lost its antigens when the primary tumor is negative. As in our study, DElia et found that in seven of eight patients, the normal appearing epithelium showed a greater density of blood group substances than did the corresponding tumor tissue. One of the patients of type A blood expressed H antigen on the urothelium. The A, B, and H blood group antigens are carbohydrate structures attached to both glycolipids and glycoproteins on the cell surface. The H structure, an oligosaccharide, is the precursor for the other blood group antigens. Whether a person is type A or B depends on whether N-acetylgalactosamine or galactose is added to the H structure. Cuadrado and co-workers26 investi- gated the expression of H antigen in a series of 89 patients with blood types A or B. He found that patients frequently expressed H after they lost the expression of A or B. Stell- ner et a~* ' suggested that the preservation of H antigen was due to sequential deletion of the transferases necessary in the production of the blood group antigens. Fujita et al.23 claimed that the deletion of the H precursor was of greater prognostic significance than the loss of the A or B antigen. There is evidence that many of the molecules identified as tumor-associated antigens are in fact car- bohydrate structures that hold the blood group antigen onto the cell surface.28 The patient in this study who ex- pressed the H precursor did not, in fact, respond to in- travesical chemotherapy. Walton and associates2' showed B2M to be a marker of malignant transformation in transitional cell carci- noma, and Adams et aL2' have demonstrated the com- plementary nature of B2M and ABO(H) antigens as pre- dictors of aggressive tumor behavior. The ABO(H) anti- gens are lost first from the apical surface of the urothelium and retained longest in the basilar layer. B2M, on the other hand, is lost first from the basilar layer. All but one patient in this study followed this previously described pattern.*' This patient was blood type O(H), had both apical and basilar staining for H antigen in one section of urothelium, and responded to intravesical chemo- therapy. One problem with using ABO(H) blood group antigens on the urothelium for prognosis is that approximately 22% of the population are nonsecretors of blood group antigens. They lack H-transferase, the enzyme responsible for the conversion of precursor substances to the H-an- 3028 CANCER Jzirzc 15 1991 Vol. 67 tigen. The urothelium in this population demonstrates decreased amount of A antigen and no H antigen.” In light of this, the interpretation of a failure to stain for H antigen as foreshadowing aggressive behavior of a lesion may be a misinterpretation approximately 20% of the time. One advantage of the combined use of ABO(H) antigens and B2M for prognostication is that the entire population expresses B2M (i.e., there are no nonsecretors of B2M). Additionally. there were no false-negative find- ings for B2M in this study; all patients whose tumors lacked B2M responded poorly to chemotherapy. For these reasons, staining for the expression of B2M in addition to ABO(H) antigen increases the predictive value of epi- thelial antigen markers. We examined adjacent normal urothelium for normal antigen staining to confirm the positive secretor status of each patient. Only one tumor of the nine which failed to stain for ABO(H) lacked nor- mally staining adjacent urothelium and he was therefore designated a nonsecretor. This tumor, in addition, lacked B2M and recurred within 2 years. Even with a small number of patients, the correlation between the presence of both B2M and ABO(H) blood group antigens with response to chemotherapy appears significant. The loss of ABO(H) antigens is a more sen- sitive predictor of recurrence, and the loss of B2M is a more specific predictor of recurrence of tumor after in- travesical BCG. The predictive value of the markers is independent of histologic grade of the tumor. The com- bination of the two markers yields a greater prognostic accuracy than either marker used alone in predicting the efficacy of intravesical treatments for superficial bladder cancer. REFERENCES 1. Cutler SJ, Heney NM, Friedell GH. Longitudinal study of patients with bladder cancer: Factors associated with disease recurrence and pro- gression. In: Bonney w , Prout G, eds. AUA Monographs, vol. 1. Bladder Cancer. Baltimore: William & Wilkins. 1982: 35. 2. Veenema RJ et ul. Bladder carcinoma treated by direct instillation of thiotepa. Urology 1965; 2:38 I . 3. National Bladder Cancer Collaborativc Group A. The role of in- travesical thiotepa in the management of superficial bladder cancer. Cuncer Res 1977; 37:2916. 4. Modes A, Eidinger D. Bruce AW. Intracavitary bacillus Calmette- Guerin in the treatment of superficial bladder tumors. J Urol 1976; 116: 180. 5. Brosman SA. Experience with bacillus Calmette-Guerin in patients with superficial bladder carcinoma. J Urol 1982; 12827. 6. Schellhammer PF, Ladaga LE, Fillion MB. Bacillus Calmette- Guerin for superficial transitional cell carcinoma of the bladder. J Urol 1986; 135(2):261. 7. Mydlo JH, Usher SM, Camacho F, Freed S. Retrospective study of efficacy of intravesical BCG alone in treatment of superficial bladder cancer. Uro1og.v 1986; 28(3): 173. 8. Lamm DL, Thor DE, Stogdill VD. Radwin HM. Bladder cancer immunotherapy. J Urol 1982; 128:931. 9. Kelley DR. Ratliff TL, Catalona WJ et (11. Intravesical bacillus Calmette-Guerin therapy for superficial bladder cancer: Effect of bacillus Calmette-Guerin viability on treatment results. J Urol 1985; I34:48. 10. Bergman SS. Javadpour N . The cell surface antigen A, B, or O(H) as an indicator of malignant potential in Stage A bladder carcinoma: Preliminary report. J Urol 1978: I19:49. I 1. Lange PH, Limas C, Fraley EE. Tissue blood-group antigens and prognosis in low stage transitional cell carcinoma of the bladder. J Urcil 1978; 11952. 12. Emmott RC, Javadpour N. Bergman SM. Soares T. Correlation of the cell surface antigens with stage and grade in cancer of the bladder. J Urol 1979; 121:37. 13. D’Elia FL, Cooper HS, Mulholland SG. Isoantigens in Stage 0 papillary transitional cell carcinoma of the bladder: Correlation with biological behavior. J Urol 1982; 127:665. 14. Davidsohn I. Stejskal R, Lill P. The loss of isoantigens A, B, and H in carcinoma of the urinary bladder. Lub Invest 1973; 28:382. 15. Bishop MC. Blood group antigens and bladder cancer. Br J Med 1982: 284:1426. 16. Coon JS, McCall A, Miller AW. Farrow GM, Weinstein RS. Expression of blood-group antigens in carcinoma in siric of the urinary bladder. Cancer 1985; 56:797. 17. DeCenzo JM, Howard P, Irish E. Antigenic deletion and prognosis ofpatients with Stage A transitional cell bladder carcinoma. J Uroll975; 114374. 18. Juhl BA, Hartzen SH. Hainau 6. Lewis A antigen in transitional cell tumors of the urinary bladder. Cancer 1986; 58:222-228. 19. Coon JS. Weinstein RS, Summers JL. Blood group precursor T- antigen expression in human urinary bladder cancer. A n J c‘liti Purhol 1982: 77596-699. 20. Walton GR, McCue PA. Graham SD. Beta-2 microglobulins as a differentiation marker in bladder cancer. J Urol 1986; 136:l 197. 2 I . Coon JS, Weinstein RS. Detection of ABH tissue isoantigens by immunoperoxidase methods in normal and neoplastic urothelium: Comparison with the erythrocyte adherence method. Am J Clin Puthol 1981; 76:163. 22. Boileau MA, Cowles RS, Schmidt KL, Schmidt WA. Comparison of specific red-cell adherence and immunoperoxidase staining techniques for ABO( H) blood-group cell-surface antigens on superficial transitional cell carcinoma of the bladder. J Sirrg Oncol 1985; 30:72. 23. Fujita J, Matsumoto K, Kishi K. Ishiyama I. Synthesis of ABH blood group substances in bladder tumors. Br J Urol 1981: 52:44. 24. Festenstein H, Schmidt W. Variation in MHC antigenic profiles of tumor cells and its biological effects. Imniirnol Rev 198 I ; 60%. 25. Stein BS. Specific red cell adherence: Immunologic evaluation of random mucosal biopsies in carcinoma of the urinary bladder. J Urol 1981; 126:37. 26. Cuadrado E, Rodriguez-Trinidad A, Blasco E, Torrado J, Garcia JAL, Arozena F. Blood group isoantigens ABO (H) in transitional car- cinoma ofthe bladder: A clinicopathological study. J Urol 1986; 135(2): 409. 27. Stellner K , Hakomori S. Warner GA. Enzymatic conversion of H glycolipid to A or B glycolipid and deficiency of these enzyme activities in adenocarcinoma. Biochim Biophys Res Curnrnitn 1973; 55:439. 28. Feizi T. Demonstration by monoclonal antibodies that carbo- hydrate structures of glycoproteins and glycolipids are oncodevelopmental antigens. Nutnre 1985; 314:53. 29. Adams J, McCue P, Graham S. Manuscript in preparation. 30. Juhl BA, Hartzen SH, Hainau B. A. B, H antigen expression in transitional carcinomas of the urinary bladder. Cancer 1986; 57: 1768. Orihuela E, Varadachay S, Herr HW, Melamed MR. Whitmore WF. The practical use of tumor marker determination in bladder washing specimens. Cuncer 1987; 60:1009. Flanigan RC, King CT. Clark TD et ul. Immunohistochemical dem- onstration of blood group antigens in neoplastic and normal human urothelium: A comparison with standard red cell adherence. J Urol 1983; 1301499-503.