h t . J. Cancer (Pred. Oncol,): 69,170-173 (1996) 0 1996 Wiley-Liss, Inc. (* &- - . Publication of the International Union Against Cancer Publication de I'UnionInternationale Contre le Cancer IMMUNOHISTOCHEMICAL DETECTION OF ADHESION MOLECULE CD44 SPLICE VARIANTS IN LYMPH NODE METASTASES OF CERVICAL CANCER Christian KAINZ~.~, Clemens TEMPFER', Petra KOHLBERGER', Seline JANISCHI, Heinz KOELBL',Gerald GITSCH' and Gerhard BREITENECKER~ 'Departntent of Gynecology and Obstetricsand ?Instituteof Pathology, Gynecopathological Unit, University of Vienna Medical School, Kenna, Austria. Expression of specific cell adhesion molecule CD44 isoforms (splice variants) has been shown to be associated with poor prognosis in human cervical cancer. We used 3 different variant exon sequence-specific murine monoclonal antibodies (MAbs) to epitopes encoded by exons v5, v6 and v7-v8 of human variant CD44 to study the expression of CD44 splice variants in 35 primary squamous-cell carcinomas of the cervix and pelvic lymph node metastases by means of immunohistochemistry. Primary tumors showed expression of CD44 splice variants CD44v5, CD44v6 and CD44v7-8 in 93%, 73% and 33% of cases, respectively. Lymph node metastases expressed CD44v5, CD44v6 and CD44v7-8 in 83%, 53% and 2 I YOof cases, respectively. Tumors with expression of CD44v6 in pelvic lymph node metastases showed metastaticspread to 2 or more pelvic lymph nodes significantly more often compared to patients without expression of splice variant CD44v6. Patients suffering from tumors with lymph node metastases expressing splice variant CD44v6 had a poorer recurrence-free survival compared to patients without CD44v6 expression in lymph node metastases, but this trend was not statistically significant. Expression of CD44 splice variants containing epitopes encoded by exon v6 in primary tumors and pelvic lymph node metastases of cervical cancer patients is consistent with a prominent role of CD44 in the process of metastasisformation. o 1996 Wiley-Liss,Inc. The transmembrane receptor protein CD44 belongs to the family of cell surface adhesion molecules involved in cell-cell and cell-matrix interactions (Jalkanen et al., 1987). CD44 mediates lymphocytic functions such as cell activation, motility, division, adhesion to the extracellular matrix and adhesion to stromal cells (Huet et al., 1989; Stamenkovic et al., 1991). CD44 expression is postulated to control lymphocytic recirculation by mediating lymphocytic binding to peripheral lymph nodes and mucosal and synovial endothelial cells (Jalkanen et al., 1987). Various human malignancies express isoforms of CD44, characterized by the insertion of additional amino acids in the extracellular domain of the CD44 protein (Gunthert et a/., 1991; Screaton et al., 1992). CD44 proteins are encoded by a gene located on chromosome 11. By post-translational modifications of pre-messenger RNA, i.e., alternative splicing, numerous isoforms of the CD44 protein are produced (CD44 splice variants CD44vl-CD44v10) (Screaton et a/., 1992; Smith et a/., 1989; Matsumura and Tarin, 1992). Aberrant expression of CD44 splice variants, eg., CD44 splice variants V6 and V7-8. in human tumors indicates a loss of splice control in malignantly transformed cells (Heider et al., 1993) and has been shown to be associated with metastasis and poor prognosis in breast cancer, colorectal cancer and gastro-intestinal lymphoma (Kaufmann et a/., 1995; Wielenga et a/., 1993; Joensuu et a/., 1993). In previous studies, we showed that in cervical cancer over-expression of CD44v6 is associated with poor prognosis (Kainz et a/., 1995~).In cervical cancer stage 111, over-expression of CD44v6 and CD44v7-8 is correlated with unfavorable prognosis (Kainz et al., 1995b). Considering the physiological function of CD44, it has been speculated that malignant cells expressing CD44 splice variants are mimicking lymphocytes and thus manage to perform the cell-cell and cell-matrix interactions necessary for metastatic spread (Dall et al., 1994; Herrlich et al., 1993~).An analogous expression pattern of CD44 variants in primary tumors and regional lymph nodes would support the hypothesis that aberrant expression of CD44 adhesion molecules on tumor cells is a crucial condition of their capability to form metastases. Another point of interest is whether expression of CD44 variants can be used to identify patients with a better prognosis within the high-risk collective of lymph nodepositive patients. To address these questions we examined the expression of CD44 splice variants CD44v5, CD44v6 and CD44v7-8 in primary tumor and lymph node samples of 35 patients with surgically treated cervical cancer stages IB to IIB. PATIENTS AND METHODS Patients Thirty-five patients with surgically treated squamous-cell carcinoma (SCC) of the cervix and metastatic disease in the pelvic lymph nodes were included in the study. In all women diagnosis was established pre-operatively by biopsy or conisation. All patients were treated by radical hysterectomy and pelvic lymphadenectomy between 1984 and 1990 and underwent post-operative adjuvant radiotherapy. Staging was done according to UICC criteria. Stages IB, IIA and IIB were present in 14, 5 and 16 cases, respectively. Histologically, tumors were graded as well (Gl, n = l l ) , moderately (G2, n = 16) and poorly (G3, n = 8) differentiated. All patients underwent a close follow-up scheme consisting of regular visits at 3-month intervals, including clinical examination with vagino-rectal palpation and serology with tumor marker evaluation. Ultrasound examination of the pelvis and chest X-ray were performed every 6 months. In cases of suspect clinical findings or when serology showed elevated marker levels, computed tomography was performed. The follow-up period ranged between 3 and 8 years (median length 51.7 months). From each lymph node specimen three different slides with tumor cells present were chosen for staining. From the primary tumor specimens, 1 slide representing the bulk of the primary tumor was chosen for staining. For CD44 splice variant analysis, we used 3 different variant exon sequence-specific murine monoclonal antibodies (MAbs) to epitopes encoded by exons v5, v6 and v7-v8 of human variant CD44. lmmunohistochemistly We prepared all sections from routinely formalin-fixed and paraffin-embedded surgery samples to evaluate the expression whom correspondence and reprint requests should be sent, at Department of Gynecology and Obstetrics, Vienna University Medical School, A-1090 Vienna, Spitalg. 23, Austria. Fax: (43) 1-40400-2907. Received: November 8,1995 and in revised form January 22,1996. CD44 IN CERVICAL CANCER of epitopes encoded by CD44 splice variants CD44v5, CD44v6 and CD44v7-8. Sections were deparaffinized in xylene and rehydrated in a graded alcohol series (100% to 70%). To recover antigenicity, we used the Antigen Retrieval System (BioGenex, San Ramon, CA), twice for 20 min in a microwave at 600 W power (HM 146, Elektra Bregenz, Schwaz, Austria), then rinsed sections in 10 mM PBS (pH 7.6). Three different variant exon sequence-specific murine MAbs were used: the first was a MAb specific for the epitope encoded by exon v5 of human variant CD44 (CD44v5, clone VFF-8; Bender, Vienna, Austria), the second MAb was specific for the epitope encoded by exon v6 of human variant CD44 (CD44v6, clone VFF-7; Bender) and the third MAb was specific for the epitope encoded by exons v7-8 of human variant CD44 (CD44v7-8, clone VFF-17; Bender). The primary antibody was diluted in serum/PBS and sections incubated for 60 min and then for a further 30 min with biotinylated anti-mouse and anti-rabbit link antibody (LSAB 2 Kit, DAKO, Carpinteria, CA). After rinsing in PBS, sections were coated with streptavidin conjugated to alkaline phosphatase for 10 min. Sections were then rinsed in PBS, incubated with Fast red chromogen (naphtol phosphate substrate in Tris buffer, Fast red chromogen tablets 5 mg; BioGenex) and then washed with distilled water. Sections were finally counter-stained with hematoxylin and mounted. Staining was judged as positive when ( i ) more than 20% of the tumor area showed strong membrane staining or (ii) more than 80% of the tumor area showed weak but unequivocal membrane staining. All other cases were judged as being negative. Positive control. The positive control slide was prepared from epidermal tissue, known to contain the antigen. In the positive control tissue, all MAbs were stained similarly. Negative control. The negative control slide was prepared from the same tissue block as the specimen. Instead of the primary antibody we used a normal, non-immune serum supernatant from the same source as the primary antibody. Statistics The x2 test was used where appropriate. Results were analyzed for the end-point of overall and recurrence-free survival. Overall survival was defined as the period between the date of primary surgery and death. Survival times of patients still alive were censored with the last follow-up date (last follow-up date or date of death), respectively. Survival probabilities were calculated by the product limit method of Kaplan and Meier (1958). Differences between groups were tested using the Wilcoxon test. All p values are results of 2-sided tests. The BMDP statistical software system (BMDP Statistical Software, Los Angeles, CA) was used for calculations. 171 FIGURE 1- Expression of CD44-containing epitope v6 in invasive cervical carcinoma tissue (white arrows); cervical stroma cells are negative. Carcinoma in situ sections are seen as well (black arrows). Bar = 40 km, counter-stain, hematoxylin. FIGURE 2 - Expression of CD44-containingepito e v6 in metastatic tumor cells in a pelvic lymph node specimen. 8D44v6 is seen on the cell membrane of tumor cells. Bar = 20 km, counter-stain, hematoxylin. shows expression of CD44v6 on the cell membranes of a cervical carcinoma specimen. We examined only invasive carcinomas. However, areas of carcinoma in situ were present in some areas, as shown in Figure 1. In these cases, there was no difference in CD44 expression between the invasive section of the tumor and the carcinoma in situ (Fig. 1). Figure 2 shows expression of CD44v6 on the cell membranes of metastatic RESULTS tumor cells in a pelvic lymph node specimen. The expression. patterns of CD44 splice variants in primary tumors and Median age of the patients was 43.4 (SD 11.6, minimum 26, corresponding lymph node metastases are shown in Table I. maximum 71) years. Mean number of nodes removed during Tumors with expression of CD44v6 in pelvic lymph node pelvic lymphadenectomy was 16.2 (8 minimum, 35 maximum). metastases showed metastatic spread to 2 or more pelvic lymph Primary tumors showed expression of CD44 splice variants CD44v5, CD44v6 and CD44v7-8 in 93%, 73% and 33% of nodes significantly more often compared with patients without cases, respectively. There was no correlation of CD44 expres- expression of splice variant CD44v6 (59% vs. 23%, x2 test, sion to tumor stage or grading. Lymph node metastases p = 0.05). Expression of splice variants CD44v5 and CD44v7-8 expressed CD44v5, CD44v6 and CD44v7-8 in 83%, 53% and in pelvic lymph node metastasis showed no correlation with the 21% of cases, respectively. We found homogenous staining in extent of lymph node metastasis. Expression of the 3 examined CD44 splice variants in the tumors and lymph nodes that were considered positive for CD44 expression. In all positive cases, we found a membrane primary tumor was of no prognostic value. Patients suffering staining pattern. No cytoplasmic staining was present. Figure 1 from tumors with lymph node metastases expressing splice KAINZ ETAL. 172 TABLE I - EXPRESSION PATTERN OF CDJ4 SPLICE VARIANTS CD44v5, CD44v6 AND CD44v7-8 IN PRIMARY TUMORS AND CORRESPONDING MFTASTATIC LYMPH NODES ~ Tumor-uositive, metastasispositive Tumor-positive, metastasisnegative Tumor-negative, metastasisnegative Tumor-negative, metastasispositive 1 .=: 1.0 3 8 CD44y5 expression CD44y6 expression CD4Jvl-8 expresston + 83% 53% 21% C 10% 20% 12% 7% 277% 67% 0% 0% 0% e 0 variant CD44v6 had a poorer recurrence-free survival compared with patients without CD44v6 expression in lymph node metastases (Wilcoxon test, p = 0.07), but this trend was not statistically significant. Kaplan-Meier analysis regarding recurrence-free survival of patients with and without CD44v6 expression in pelvic lymph node metastases is shown in Figure 3. Overall survival was not significantly associated with CD44v5, CD44v6 or CD44v7-8 expression in lymph node metastases (Wilcoxon test, p = 0.1, p = 0.09, p = 0.4, respectively). E9 n 0.5 I I I '1I No CD44 v6Ixpression CD44 v6-Expression Q g 0 20 40 60 80 100 120 140 Months Since Initial Treatment FIGURE3 - Overall survival of patients suffering from tumors with or without expression of splice variant CD44v6 in pelvic lymph node metastases (Wilcoxon test,p = 0.07). DISCUSSION We investigated 35 patients with cervical cancer and metastatic spread to the regional lymph nodes to learn more about the role of the cell adhesion molecule CD44 in tumor metastasis. Others have hypothesized that expression of the cell adhesion molecule CD44 plays a crucial role in the natural history of the metastatic process (Dall et al., 1994; Stamenkovic et al., 1991). In a previous study, we supported this hypothesis by finding a significant correlation between expression of CD44 splice variants and the probability of metastasis in cervical cancer patients (Kainz et al., 199%). Expression of CD44 in lymph node metastasis of cervical cancer, however, has not been investigated. We found that CD44 splice variants are expressed in metastatic tumor cells of cervical cancer patients. Given the hypothesis that a cell clone with high expression of CD44v6 would possess a selection advantage in the process of metastasis formation (Herrlich et al., 1993a, b ) , we expected a higher rate of CD44v6 expression in metastatic lymph nodes compared with primary tumor. However, we found a lower expreTsion rate in the metastatic lymph nodes for all 3 splice variants of CD44 investigated (Table I). A possible explanation for CD44 down-regulation could be found in the influence of a different organic environment altering the expression pattern of cell surface molecules in regional lymph nodes. Shedding of membrane-bound molecules after having accomplished their function could also account for a weaker staining in regional lymph nodes compared with primary tumor (Dolo et al., 1994; Shibata et al., 1993). In a previous study, we found a prognostic value for CD44v6 expression in the primary tumor of patients free of pelvic lymph node metastasis. Contrary to that finding, expression of splice variants CD44v5, CD44v6 and CD44v7-8 in primary tumor failed to predict a poor outcome in patients displaying metastatic disease in pelvic lymph nodes (Kainz et al., 1995~). Such a high-risk collective, consisting only of patients with histologically proven metastatic spread to the regional lymph nodes, was further investigdted in the present study. As expected, expression of CD44v6 in primary tumor was not predictive of outcome. We could show that expression of CD44v6 in the metastatic tumor cells of pelvic lymph nodes is associated with a trend to poorer prognosis (Fig. 1). If our results are confirmed in a larger series of patients, expression of CD44v6 in pelvic lymph node metastasis could be used as a means to identify patients with a favorable prognosis within the high-risk collective of node-positive patients. In a previous study, we also showed that the incidence of pelvic lymph node metastasis is significantly higher when primary tumors express CD44v6 (Kainz et a/., 1 9 9 5 ~ )In . the present study, a significantly higher number of positive lymph nodes was found when CD44v6 expression was seen in metastatic lymph nodes. These results underline again the functional role of CD44v6-expressing cells in the expansion of cervical cancer beyond regional limits. Although the expression rate of CD44 splice variants in metastatic lymph nodes was lower than in primary tumor, we found that CD44v6 expression in pelvic lymph node metastases is associated with a trend to poorer prognosis and that it is predictive of the extent of pelvic lymph node metastasis. We conclude that tumors which conserve their capability of CD44v6 expression in the altered environment of pelvic lymph nodes have the worst prognosis. We showed that CD44 splice variants are expressed in pelvic lymph node metastasis. Expression of CD44v6 in lymph node metastases of cervical cancer patients is a promising marker for identifying patients with a favorable prognosis within the high-risk collective of node-positive patients. The functional role of CD44 splice variant expression in the natural history of metastasis is underlined by the correlation between CD44v6 expression and the extent of pelvic lymph node involvement in cervical cancer patients. ACKNOWLEDGEMENTS This work was supported by a research grant from the mayor of Vienna (1045) to C.K. CD44 IN CERVICAL CANCER 173 REFERENCES DALL,P., HEIDER, K.H., HEKELE. A,. VON MINCKWITZ, G., KAUFMA", KAINZ,C., KOHLBERGER, P., SLIUTZ,G., TEMPFER, C., HEINZL,H., M., PONTA,H. and HERRLICH. P., Surface protein expression and REINTHALLER, R., BREITENECKER, G. and KOELBL, H., Splice variants messenger RNA-splicing analysis of CD44 in uterine cervical cancer of CD44 in human cervical cancer stage IB to IIB. Gynecol. Oncol., 57, and normal epithelium. Cancer Res., 54,3337-3341 (1994). 383-387 (1995a). P., TEMPFER, C., SLIUTZ,G., GITSCH,G., DOLO,V., GINESTRA, A,. GHERSI,G., NAGASE,H. and VITTORELLI, KAINZ, C., KOHLBERGER, A. and BREITENECKER, G., Prognostic value of CD44 M.L., Human breast carcinoma cells cultured in the presence of serum REINTHALLER, shed membrane vesicles rich in gelatinolytic activities. J. Submicrosc. splice variants in human cervical cancer stage 111. Europ. J. Cancer, 31, 1706-1709 (199%). Cytol. Pathol., 26,173-180 (1994). E.L. and MEIER,P., Nonparametric estimation from incomGUNTHERT, U., HOFMANN, M.. RUDY,W., REBER,S., ZOLLER,M., KAPLAN, HAUSSMANN, I., MATZKU, S., W E N Z ~A,. L , PONTA,H. and HERRLICH, plete observations. J. Amer. Stat. Assoc., 53,457481 (1958). P., A new variant of glycoprotein CD44 confers metastatic potential to KAUFMANN, M., HEIDER,K.H., SINN,H.P., VON MINCKWITZ, G., rat carcinoma cells. Cell, 65, 13-24 (1991). PONTA,H. and HERRLICH, P., CD44 variant exon epitopes in primaly HEIDER,K.H., DAMMRICH, J.. SKROCHANGEL.P., MULLERHER- breast cancer and length of survival. Lancer, 345,615-619 (1995). MELINK,H.K., VOLLMERS, H.P., HERRLICH, P. and PONTA,H., MATSUMURA, Y. and TARIN,D., Significance of CD44 gene products Differential expression of CD44 splice variants in intestinal- and for cancer diagnosis and disease evaluation. Lancet, 340, 1053-1058 diffuse-type human gastric carcinomas and normal gastric mucosa. (1992). CancerRes., 53,4197-4203 (1993). SCREATON, G.R., BELL,M.V., JACKSON, D.G., CORNELIS, F.B., GERTH, HERRLICH, P., ZOLLER,M.. PALS,S.T. and PONTA,H., CD44 splice U. and BELL,J.I., Genomic structure of DNA encoding the lymphovariants: metastases meet lymphocytes. Immunol. Today. 14, 395-399 cyte homing receptor CD44 reveals at least 12 alternatively spliced (1993~). exons. Proc. nut. Acad. Sci. (Wash.), 89,12160-12164 (1992). HERRLICH, P., ZOLLER,M., PALS,S.T. and PONTA,H., CD44 splice SHIBATA, T., MICALLEF, M., CHIBA, I., HOSOKAWA, M. and KOBAYASHI, variants: metastases meet lymphocytes. Immunol. Toda.y, 14, 395-399 H., Modulation of the shedding of a rat tumor-associated antigen by (1993b). growth regulation and anti-cancer drugs. Anticancer Drugs, 4,657-664 HUET,S.. GROUX, H., CAILLOU, B., VALENTIN, H., PRIEUR. A.M. and (1993). BERNARD, A,, CD44 contributes to T cell activation. J. Immunol., 143, SMITH,C.W., PATTON,J.G. and NADALGINARD,B., Alternative 798-801 (1989). splicing in the control of gene expression. Ann. Rev. Genet., 23, JALKANEN, S., BARGATZE, R.F., DE LOS Touos, J. and BUTCHER, E.C., 527-577 (1989). Lymphocyte recognition of high endothelium: antibodies to distinct STAMENKOVIC, I., ARUFFO,A., AMIOT,M. and SEED,B., The hematoepitopes of an 85-95-kD glycoprotein antigen differentially inhibit poietic and epithelial forms of CD44 are distinct polypeptides with lymphocyte binding to lymph node, mucosal, or synovial endothelial different adhesion potentials for hyaluronate-bearing cells. EMBO J., cells. J. Cell Eiol., 105,983-990 (1987). 10,343-348 (1991). JOENSUU, H., RISTAMAKI, R., KLEMI,P.J. and JALKANEN, S., Lympho- WIELENGA, V.J.M., HEIDER,K.H., OFFERHAUS, J.A., ADOLF,G.R., cyte homing receptor (CD44) expression is associated with poor VAN DEN BERG, F.M.. PONTA,H., HERRLICH, P. and PALS,S.T., prognosis in gastrointestinal lymphoma. Brit. J. Cancer, 68, 428-432 Expression of CD44 variant proteins in human colorectal cancer is (1993). related to tumor progression. Cancer Res., 53,47544756 (1993).