Neural cell adhesion molecule expression in renal cell carcinomas: relation to metastatic behavior

Original Contributions Neural Cell Adhesion Molecule Expression in Renal Cell Carcinomas: Relation to Metastatic Behavior LAURENT DANIEL, MD, PHD, CORINNE BOUVIER, MD, BRUNO CHETAILLE, MD, JOANNY GOUVERNET, MD, ALINE LUCCIONI, MD, DOMINIQUE ROSSI, MD, ERIC LECHEVALLIER, MD, XAVIER MURACCIOLE, MD, PHD, CHRISTIAN COULANGE, MD, AND DOMINIQUE FIGARELLA-BRANGER, MD, PHD Neural cell adhesion molecule (NCAM), a member of the immunoglobulin superfamily, is expressed by a subgroup of renal cell carcinomas (RCCs) and by a limited number of adult organs, including the central nervous system (CNS) and adrenal gland. Because the major function of NCAM is homophilic adhesion between homotypic and heterotypic cells, we hypothesized that NCAM-expressing RCCs should preferentially metastasize to the CNS and adrenal gland. We did a retrospective immunohistochemical analysis of NCAM expression both in 338 primary renal tumors, including 249 conventional RCCs and 31 metastases of conventional RCCs. In primary renal tumors, NCAM was expressed by only 38 (15.2%) conventional RCCs and by no other histological subtypes of renal tumor. This expression correlated with a higher risk of adrenal and CNS metastases (P <0.001). NCAM expres-

sion also correlated with tumor size (P <0.001), renal vein involvement (P ⴝ 0.02), perirenal invasion (P ⴝ 0.02), and Fuhrman grading (P < 0.001). Finally, patients with NCAM-expressing RCCs had a lower survival rate (P ⴝ 0.006), especially in the first 2 years after surgery. NCAM expression is of interest both for evaluating the prognosis of patients with conventional RCCs and for determining a subgroup of patients at high risk for adrenal and CNS metastases. HUM PATHOL 34:528-532. © 2003 Elsevier Inc. All rights reserved. Key words: renal tumor, neural cell adhesion molecule, metastasis. Abbreviations: CNS, central nervous system, cRCC, conventional renal cell carcinoma; NCAM, neural cell adhesion molecule, RCC, renal cell carcinoma; pRCC, papillary renal cell carcinoma, PSA, polysialylated.

Cell adhesion molecules, including the immunoglobulin superfamily, play a crucial role in controlling tumor development and the metastatic cascade.1,2 We recently showed that the polysialylated (PSA) isoforms of the neural cell adhesion molecule (NCAM) are involved in the metastatic process by promoting cell dissociation.3,4 Nevertheless, the role of the nonsialylated NCAM in the metastatic process has not been established in most carcinomas. In kidneys, NCAM is expressed during development,5 postischemic regeneration,6 and formation of a subgroup of renal cell carcinomas (RCCs).7,8 NCAM is also expressed in rare benign tumors originating from the distal part of the nephron.9 Because NCAM is highly expressed in the central nervous system (CNS)10 and adrenal gland,11 which are common sites of RCC metastases,12,13 we hypothesized that NCAM-expressing

RCCs should preferentially metastasize into these tissues according to the major function of NCAM, that is, homophilic adhesion NCAM-NCAM. Hence, we retrospectivelly analyzed NCAM expression in primary renal tumors and in metastases of RCC. Results showed that NCAM-expressing RCCs behaved aggressively and metastasized preferentially to NCAMexpressing organs like the CNS and adrenal gland.

From the Departments of Pathology, Statistics, and Radiotherapy, Centre Hospitalier Universitaire (CHU) Timone, Marseille, France; and the Department of Urology, North and Salvator Hospitals, Marseille, France. Accepted for publication February 12, 2003. Supported by the Conseil Ge´ne´ral des Bouches-du-Rhoˆne. Address correspondence and reprint requests to Laurent Daniel, MD, PhD, Department of Pathology, CHU Timone, 264 rue St.Pierre, 13385 Marseille, Cedex 05, France. © 2003 Elsevier Inc. All rights reserved. 0046-8177/03/3406-0008$30.00/0 doi:10.1016/S0046-8177(03)00178-3

MATERIALS AND METHODS Patients and Tumor Samples A total of 338 patients who underwent nephrectomy or percutaneous biopsy for renal tumors in our institution between January 1989 and January 1999 were consecutively selected. Indications for a protocol biopsy were as described elsewhere.14 All patients underwent a yearly imaging evaluation including at least hepatic ultrasonography and computed tomography of the lung. Cerebral computed tomography was performed if clinical manifestations were present. The mean follow-up period was 46.7 months ⫾ 36.2 (range, 1 to 143.9 months). The follow-up data were recorded for all patients except those with oncocytoma, a well-established benign tumor.15

Histological Classification Gross examination of nephrectomy specimens was done using the TNM staging system developed by the International

528

NEURAL CELL ADHESION MOLECULE EXPRESSION IN RENAL CELL CARCINOMAS (Daniel et al)

FIGURE 1. (A) Adrenal gland NCAM expression was strong in the medulla (arrowhead) and mild in the outer layers of the adrenal cortex (arrow). (Original magnification ⫻ 40.) (B) The normal adult kidney showed only weak NCAM expression in a few collecting ducts. (Original magnification ⫻ 100.) (C) Grade 1 NCAM immunostaining in a cRCC. Note that immunoreactive cells are at the periphery of the tumor. (Original magnification ⫻ 100.) (D) Grade 2 NCAM immunostaining in a cRCC with a tubulomicrocystic pattern. (Original magnification ⫻ 250.) (E) Grade 3 NCAM immunostaining in a cRCC. All of the tumor cells exhibited strong staining. (Original magnification ⫻ 100.) (F) Cerebral metastasis of a cRCC. All tumor cells strongly expressed NCAM and were intermingled with the brain matrix (asterisk). (Original magnification ⫻ 400.)

Union Against Cancer and the American Joint Committee on Cancer.16 Renal tumors were histologically diagnosed according to the international classification.17 Histological tumor grading was applied according to the classification system of Fuhrman.18 This grading was done for conventional RCCs (cRCCs) and papillary RCCs (pRCCs) only.

Immunohistochemistry Studies Immunohistochemistry was performed on 5-␮m-thick serial paraffin-embedded sections from specimens previously fixed in 10% buffered formalin. Antigen retrieval was performed 3 times for 5 minutes each for each section (750-W

529

HUMAN PATHOLOGY

TABLE 1.

Volume 34, No. 6 (June 2003)

Metastatic Occurrence at the End of Follow-Up

NCAM expression in primary renal tumors

No metastasis

Occurrence in NCAM-lacking organs

Occurrence in NCAM-expressing organs

Total

NCAM ⫺ NCAM ⫹ Total

146 (78.1%) 11 (30.6%) 157 (70.4%)

38 (20.3%) 8 (22.2%) 46 (20.6%)

3 (1.6%) 17 (47.2%) 20 (9%)

187 (100%) 36 (100%) 223 (100%)

NOTE: P ⬍ 0.001.

microwave; pH 6.3 buffer). The immunoperoxidase technique was performed with an automatic immunostaining device and Ventana kits (Ventana, Strasbourg, France). Sections were reacted for 30 minutes with the anti-NCAM antibody (clone 1B6, dilution 1:50; Novocastra, Le Perray en Yvelines, France). This antibody recognizes the common extracellular domain of the NCAM isoforms, but is not effective on frozen sections. In this situation, we used the clone MY31 (1:10 dilution; Becton-Dickinson, San Jose, CA) for comparison with the PSA-NCAM staining. Positive controls were the nerves within the tumors and normal adrenal glands purchased by the forensic department of our institution. As negative controls, some sections were incubated either with irrelevant antibody or phosphate-buffered saline (PBS). Ten frozen specimens from primary tumors and 2 from cerebral metastases stored at ⫺80°C that were positive for NCAM were selected for PSA-NCAM immunohistochemistry. The mouse monoclonal IgM Men-B specifically recognizing the polysialylated form of NCAM (dilution 1:1000; a gift from G. Rougon, IBDM, Marseille, France) was used. Immunoperoxidase staining was performed using an avidin-biotin-peroxidase complex (ABC) system (Vector, Burlingame, CA). Immunostaining of primary tumor cells was semiquantitatively scored according to 4 grades as previously described for another cell adhesion molecule in RCCs: no reactive cell (grade 0), ⬍25% of reactive cells (grade 1), 25% to 75% of reactive cells (grade 2), and ⬎75% of reactive cells (grade 3).19 Two pathologists (L.D. and C.B.) reviewed all slides.

Statistical Analysis Fisher’s exact probability test and Student’s t test were used. Analysis of variance was used for analysis of up to 2 means. A P value ⬍0.05 was considered significant. Multivariate analysis was performed with a logistic model in an ascending stepwise mode. The P values to enter and to remove the variable in this model were 0.05 and 0.10, respectively. The candidate variables were NCAM positivity, perirenal invasion, renal vein involvement, high (3– 4) and low (1–2) Fuhrman grades, and age. Survival curves were established using the Kaplan-Meier method and the log-rank test. Data were analyzed using BMDP system 1.1 PC software (Sedona, AZ).

RESULTS Patients and Tumor Samples Patients included 239 (70.7%) men and 99 women, with a mean age of 59.5 years ⫾ 11.45 (range, 18 to 87 years). The mean tumor size was 5.7 cm ⫾ 3.01 (range, 0.5 to 18 cm). Histological subtypes were cRCC (n ⫽ 249), pRCC (n ⫽ 50), chromophobe cell renal carcinomas (n ⫽ 9), and oncocytomas (n ⫽ 30). Among the patients who underwent nephrectomy for cRCC (n ⫽ 218; 88.4%), 136 (62.4%) were stage T1–2,

36 (16.5%) were stage T3a, 43 (19.7%) were stage T3b, and 3 (1.4%) were stage T4. Only 9 cRCCs (4.1%) showed lymph node extension. The cRCCs included 78 (31.4%) high grades (Fuhrman grades 3 and 4) and 171 (68.6%) low grades (Fuhrman grades 1 and 2). The following data were available for 282 of 308 (91.6%) patients with malignant tumors. Forty-three patients (15.2%) died during follow-up. The cause of death was metastases for 25 patients (58.2%), including 23 with cRCC and 2 with pRCC. Among the patients who died of metastases, 5 (20%) had CNS metastases. Two other patients developed a controlateral cRCC, and 3 patients had retroperitoneal recurrences. Additionally, at the end of follow-up, 41 alive patients with cRCC (16.5%) had metastases. These metastases were initially diagnosed in the following organs: lung (n ⫽ 10), bone (n ⫽ 11), CNS (n ⫽ 10), liver (n ⫽ 4), adrenal gland (n ⫽ 4), and skin (n ⫽ 2). Metastatic tissue specimens (n ⫽ 31) were obtained from 8 patients of the 249 with available primary tumors. In the remaining cases, only the metastatic specimen was available because nephrectomies or biopsies were not performed according to the clinical status. Metastatic specimens were located in the following organs: bone (n ⫽ 13), CNS (n ⫽ 9), adrenal gland (n ⫽ 4), skin (n ⫽ 3), liver (n ⫽ 1), and cervical lymph nodes (n ⫽ 1). Lung metastases were not biopsied. Immunohistochemistry Studies Positive control samples in adrenal gland were strongly stained with anti-NCAM. Staining was detected both in the outer layers of the cortex and in the medulla (Fig 1A). In normal kidney, no staining was observed except in a few collecting ducts (Fig 1B). NCAM expression was observed in no oncocytomas, pRCCs, or chromophobe cell carcinomas. Of the cRCCs, 211 (84.8%) showed no staining (immunohistochemical grade 0) for anti-NCAM antibody (Fig 1C), 17 (6.8%) were immunohistochemical grade 1 (Fig 1D), 11 (4.4%) were grade 2, and only 10 (4%) were grade 3 (Fig 1E). We found that staining was only membranous and was localized in tumor areas composed of large clear cells, often at the tumor periphery. Of the 31 metastatic specimens, 16 were NCAM positive, including all of the 13 metastases located in the CNS (Fig 1F) and adrenal gland and 3 of the 13 (23%) bone metastases. In contrast with primary tumors, all metastatic tumor cells strongly expressed NCAM. Therefore,

530

NEURAL CELL ADHESION MOLECULE EXPRESSION IN RENAL CELL CARCINOMAS (Daniel et al)

96.4%), respectively, but the sensitivity and the specificity of NCAM expression for the specific metastatic occurrence in adrenal gland and CNS were 85% (62.1% to 96.8%) and 90.6% (85.8% to 94.3%), respectively. Multivariate analysis was performed to evaluate the predictive value of the various parameters in terms of metastatic risk and selective localization of the metastases (Table 2). High Fuhrman grades, extrarenal invasion, and NCAM expression were indicative of metastatic occurrence. In contrast, NCAM expression was the only parameter to determine the specific metastatic occurrence in CNS and adrenal gland (P ⬍0.001). DISCUSSION

FIGURE 2. Survival in relation to NCAM expression by primary cell tumors. Patients with NCAM-expressing cRCCs (n ⫽ 36) had a lower survival rate than those with NCAM-lacking cRCCs (n ⫽ 187) (P ⫽ 0.006).

NCAM expression was strongly related to the location of metastases (P ⬍0.001) (Table 1). PSA-NCAM was not found in frozen sections of the 10 primary cRCCs and 2 cerebral metastases that we studied. Correlation Between NCAM Expression in cRCCs and Other Parameters The NCAM-expressing cRCCs included 3 Fuhrman grade 1 (5.4% of this grade), 10 Fuhrman grade 2 (8.6% of this grade), 17 Fuhrman grade 3 (27.4% of this grade), and 8 Fuhrman grade 4 (50% of this grade). Hence, Fuhrman grading and NCAM expression were strongly correlated (P ⬍0.001). NCAM expression also correlated with tumor size (7.7 cm ⫾ 3.4 versus 5.6 cm ⫾ 2.8; P ⬍0.001), renal vein involvement (35.3% versus 16.8%; P ⫽ 0.02), and perirenal invasion (44.1% versus 23.4%; P ⫽ 0.02). Also, NCAM-expressing cRCCs more frequently showed lymph node involvement (10.5% versus 2.4%; P ⫽ 0.03). Patient survival was inversely correlated with NCAM expression (Fig 2). The mean survival for patients with NCAMexpressing cRCC was 94.4 months ⫾ 12.1. This survival differed from those of other patients (130.5 months ⫾ 3.6; P ⫽ 0.006). The sensitivity and the specificity of NCAM expression for the global metastatic occurrence were 37.9% (26.2% to 50.7%) and 93% (87.8% to TABLE 2. Step

We show in this study that NCAM is expressed by a subgroup of aggressive cRCCs, and it should serve to predict the behavior of these tumors. In particular, NCAM expression correlated strongly in univariate and multivariate analyses with the occurrence of metastases in the CNS and adrenal gland. Survival of patients with NCAM-expressing RCCs was lower than the survival of other patients. The difference was striking in the first 2 years of follow-up, in keeping with the fact that most brain metastases of RCC are observed during this period.20 We have previously shown that PSA-NCAM favors the metastatic process because of its role in cell disassociation,3,4 but NCAM is not polysialylated in cRCC, suggesting that it plays another role in these tumors involving homophilic adhesion. The metastatic occurrence in targeting organs should be determined by 3 main mechanisms: migration induced by chemokines,21 specific location by adhesion molecule interactions,22 or selective spread by growth factors.23 The high NCAM expression by metastatic RCC cells invading CNS and adrenal glands, 2 organs expressing high NCAM levels, strongly suggests that selective metastatic spreading is favored by homophilic adhesion between tumor cells and host organ cells. Similar mechanisms have been reported to explain both the unusual sites of involvement in the case of NCAM-expressing T cell lymphomas24 and the lytic bone lesions of multiple myeloma.25 Conversely, for other tumors like pancreatic adenocarcinomas, reduced levels of NCAM expression were found to correlate with increased tumor malignancy.26 This result was also observed in a transgenic mouse model of ␤-cell pancreatic carcinoma by cross-

Multivariate Analyses for Total Metastatic Occurrence and Specific Metastatic Occurrence Variable

P

Total metastatic occurrence 1 High grade (3–4) ⬍0.001 2 Perirenal invasion ⬍0.001 3 NCAM positivity ⬍0.001 Specific metastatic occurrence in adrenal gland and CNS 1 NCAM positivity ⬍0.001

531

␤

exp(␤)

95% confidence interval

1.39 1.44 1.56

4.03 4.21 4.77

1.88–8.61 1.94–9.12 1.86–12.24

4.12

61.89

13.06–293.24

HUMAN PATHOLOGY

Volume 34, No. 6 (June 2003)

ing these mice with NCAM knockout mice.27 The hypothesis was that reduced levels of NCAM could increase cell dissociation from primary tumors. The mechanism is different in cRCCs, probably because NCAM plays a role at another step of the metastatic process, the final one. The hypothesis that metastatic cells undergo a phenotypic change in NCAM-positive tissues is not probable according to that all metastatic cells were NCAM positive, suggesting a clonal selection. Moreover, molecules interacting with NCAM can also modulate homophilic NCAM adhesion. Indeed, other functions of NCAM are binding to the matrix proteins heparan sulfate proteoglycans and to the protein of the immunoglobulin superfamily L1, both of which are highly expressed in the CNS.28,29 Heparan sulfate proteoglycans and L1 are also present in kidney30 and may contribute to the increased migration through the extracellular matrix that would explain the local aggressive behavior that we observed. Finally, NCAM is able to bind to the endothelial cells.31 This should also favor metastatic invasion of targeting organs. Detection of NCAM expression in cRCCs could have important clinical implications. First, computed tomography of the brain should be done systematically during the first 2 years of follow-up. Second, if NCAM expression is proven on percutaneous biopsy before the surgery, adrenal gland should always be removed even in the case of a small tumor in the lower portion of the kidney. Third, because NCAM is down-regulated by interferon,32 adjuvant immunotherapy protocols should be evaluated in this subgroup of patients. In conclusion, prospective studies should confirm these results, but NCAM expression should be evaluated by immunohistochemistry in percutaneous biopsy and nephrectomy specimens from cRCCs. Further investigations are needed to decipher the molecular mechanisms leading to the increased risk of metastasis in these patients. Modulation of NCAM expression could represent a new therapeutic way to improve the survival of patients with NCAM-expressing cRCCs. REFERENCES 1. Koukoulis GK, Patriarca CP, Gould VE: Adhesion molecules and tumor metastasis. HUM PATHOL 29:889-892, 1998 2. Meyer T, Hart IR: Mechanisms of tumour metastasis. Eur J Cancer 34:214-221, 1998 3. Daniel L, Trouillas J, Renaud W, et al: Polysialylated-neural cell adhesion molecule expression in rat pituitary transplantable tumors (spontaneous mammotropic transplantable tumor in WistarFurth rats) is related to growth rate and malignancy. Cancer Res 60:80-85, 2000 4. Daniel L, Durbec P, Gautherot E, et al: A nude mice model of human rhabdomyosarcoma lung metastases for evaluating the role of polysialic acids in the metastatic process. Oncogene 20:997-1004, 2001 5. Nouwen EJ, Dauwe S, Van der Biest I, et al: Stage- and segment-specific expression of cell-adhesion molecules N-CAM, ACAM, and L-CAM in the kidney. Kidney Int 44:147-158, 1993 6. Abbate M, Brown D, Bonventre JV: Expression of NCAM recapitulates tubulogenic development in kidney recovering from acute ischemia. Am J Physiol 277:454-463, 1999 7. Terpe HJ, Tajrobehkar K, Gunthert U, et al: Expression of cell adhesion molecules alpha-2, alpha-5 and alpha-6 integrin, E-cadherin, N-CAM, and CD-44 in renal cell carcinomas. An immunohis-

tochemical study. Virchows Arch A (Pathol Anat Histopathol) 422: 219-224, 1993 8. Garin-Chesa P, Fellinger EJ, Huvos AG, et al: Immunohistochemical analysis of neural cell adhesion molecules-differential expression in small round cell tumors of childhood and adolescence. Am J Pathol 139:275-286, 1991 9. Daniel L, Lechevallier E, Bouvier C, et al: Adult mesoblastic nephroma. Pathol Res Pract 196:135-139, 2000 10. Jin L, Hemperly JJ, Llyod RV: Expression of neural cell adhesion molecule in normal and neoplastic tissues. Am J Pathol 138:961-969, 1991 11. Lahr G, Mayerhofer A, Bucher S, et al: Neural cell adhesion molecules in rat endocrine tissues and tumor cells: Distribution and molecular analysis. Endocrinology 132:1207-1217, 1993 12. Yamanaka K, Gohji K, Hara I, et al: Clinical study of renal cell carcinomas with brain metastasis. Int J Urol 5:124-128, 1998 13. Paul R, Mordhorst J, Bush R, et al: Adrenal sparing surgery during radical nephrectomy in patients with renal cancer: A new algorithm. J Urol 166:59-62, 2001 14. Lechevallier E, Andre´ M, Barriol D, et al: Fine-needle percutaneous biopsy of renal masses. Radiology 216:506-510, 2000 15. Amin MB, Crotty TB, Tickoo SK, et al: Renal oncocytoma: A reappraisal of morphologic features with clinicopathologic findings in 80 cases. Am J Surg Pathol 21:1-12, 1997 16. Guinan P, Sobin LH, Algaba F, et al: Staging of renal cell carcinoma: Workgroup no. 3. Cancer 80:992-993, 1997 17. Storkel S, Eble J, Adlakha K, et al: Classification of renal cell carcinoma: Workgroup no. 1. Cancer 80:987-989, 1997 18. Fu¨ hrman SA, Lasky LC, Limas C: Prognostic significance of morphologic parameters in renal cell carcinomas. Am Surg Pathol 6:655-663, 1982 19. Daniel L, Lechevallier E, Giorgi R, et al: CD44s and CD44v6 expression in localized T1-T2 conventional renal cell carcinomas. J Pathol 193:345-349, 2001 20. Maor MH, Frias AE, Oswald MJ: Palliative radiotherapy for brain metastasis in renal cell carcinoma. Cancer 62:1912-1917, 1988 21. Muller A, Homey B, Soto H, et al: Involvement of chemokine receptors in breast cancer metastasis. Nature 410:50-56, 2001 22. Woodhouse EC, Chuaqui RF, Liotta LA: General mechanisms of metastasis. Cancer 80:1529-1537, 1997 23. Guise TA, Yin JJ, Taylor SD, et al: Evidence for a causal role of parathyroid-related protein of the pathogenesis of human breast cancer–mediated osteolysis. J Clin Invest 98:1544-1549, 1996 24. Kern WF, Spier CM, Hanneman EH, et al: Neural cell adhesion molecule-positive peripheral T-cell lymphoma. A rare variant with a propensity for unusual sites of involvement. Blood 9:2432-2437, 1992 25. Ely SA, Knowles DM: Expression of CD56/neural cell adhesion molecule correlates with the presence of lytic bone lesions in multiple myeloma and distinguishes myeloma from monoclonal gammopathy of undetermined significance and lymphomas with plasmacytoid differentiation. Am J Pathol 160:1293-1299, 2002 26. Fogar P, Basso D, Pasquali C, et al: Neural cell adhesion molecule (N-CAM) in gastro-intestinal neoplasias. Anticancer Res 17:1227-1230, 1997 27. Perl AK, Dahl U, Wilgenbus P, et al: Reduced expression of neural cell adhesion molecule induces metastatic dissemination of pancreatic ␤ tumor cells. Nat Med 5:286-291, 1999 28. Margolis RK, Margolis RU: Nervous tissue proteoglycans. Experimentia 49:429-446, 1993 29. Horstkorte R, Schachner M, Magyar JP, et al: The fourth immunoglobulin-like domain of N-CAM contains a carbohydrate recognition domain for oligomannosidic glycans implicated in association with L1 and neurite outgrowth. J Cell Biol 121:1409-1421, 1993 30. Debiec H, Christensen EI, Ronco PM: The cell adhesion molecule L1 is developmentally regulated in the renal epithelium and is involved in kidney branching morphogenesis. J Cell Biol 143:2067-2079, 1998 31. Zocchi MR, Vidal M, Poggi A: Involvement of CD56/NCAM in the adhesion of human solid tumor cell lines to endothelial cells. Exp Cell Res 204:130-135, 1993 32. Geertsen R, Zenklusen R, Kamarashev J, et al: Inverse regulation of neural cell adhesion molecule (NCAM) by IFN-gamma in melanoma cell cultures established from CNS lesions. Int J Cancer 83:135-140, 1999

532