216

CANCER
CYTOPATHOLOGY

Uroplakin as a Marker for Typing Metastatic Transitional Cell Carcinoma on Fine-Needle Aspiration Specimens
Xiaowei Xu, M.D., Ph.D.1 Tung-Tien Sun, Ph.D.2 Prabodh K. Gupta, M.D.1 Paul Zhang, M.D.1 Joseph F. Nasuti, M.D.1
1

BACKGROUND. Immunohistological markers specic for a single type of epithelium

are rare. Recently, urothelium tissue-specic genes were cloned. The genes encoded a family of transmembrane proteins, uroplakins, that are expressed only in urothelial mucosa. Using uroplakin antibodies on parafn-embedded tissue, a previous study demonstrated positive staining in 66% of metastatic transitional cell carcinoma (TCC) cases and negative staining in all other tumors (including breast, ovarian, lung, and gastrointestinal carcinomas) tested. The current study addresses the diagnostic value of uroplakins in conventional ne-needle aspiration (FNA) material in establishing a diagnosis of metastatic TCC. METHODS. Representative slides from 27 FNA cases of metastatic TCC and 52 non-TCC carcinomas were collected. The avidin-biotin-peroxidase method was utilized, using polyclonal antiuroplakin as the primary antibody on 95% ethanolxed, Papanicoloau-stained direct smears. RESULTS. Twenty-ve of 27 metastatic TCC cases (93%) were found to stain positively for uroplakin with a supercial membrane/microluminal staining pattern. A few cells with diffuse membranous staining also were noted in 48% of the positive metastatic TCC cases. The supercial membrane/microluminal staining pattern was not observed in any of the non-TCC carcinomas. However, approximately 6% of these cases (3 of 52 cases) did show rare tumor cells with diffuse membranous staining. CONCLUSIONS. The application of uroplakin antibodies to 95% ethanol-xed FNA direct smears has improved the sensitivity of the antibody for metastatic TCC while maintaining a specicity comparable to that of parafn-embedded tissue. The authors believe that these antibodies have diagnostic potential in cytopathology in the evaluation of metastatic TCC. Cancer (Cancer Cytopathol) 2001;93:216 221. 2001 American Cancer Society.

Department of Pathology and Laboratory Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. Department of Dermatology, Pharmacology, and Urology, New York University Medical School, New York, New York.

Presented in part at the 48th Annual Scientic Meeting of the American Society of Cytopathology, Philadelphia, Pennsylvania, November 711, 2000. Recipient of the Warren R. Lang, M.D., Award for the best scientic paper in cytopathology. Address for reprints: Joseph F. Nasuti, M.D., Department of Pathology and Laboratory Medicine, Division of Cytopathology and Cytometry, Hospital of the University of Pennsylvania, 6 Founders Building, 3400 Spruce Street, Philadelphia, PA 19104; Fax: (215) 662-6518; E-mail: jfnasuti@ mail.med.upenn.edu Received May 31, 2000; revision received January 1, 2001; accepted January 2, 2001. 2001 American Cancer Society

KEYWORDS: uroplakin, metastatic transitional cell carcinoma, ne-needle aspiration, immunocytochemical staining.

ransitional cell carcinoma (TCC) accounts for nearly 90% of all bladder carcinomas. It is estimated that approximately 20 30% of these tumors will progress, metastasize, and lead to the death of the patient.1 Finding the primary tumor in a patient presenting with metastatic carcinoma of unknown origin is a persistent problem for both cytopathologists and surgical pathologists. Cytomorphologic criteria such as the presence of cercariform cells in ne-needle aspiration (FNA) specimens (Fig. 1) have limited diagnostic usefulness; they appear in 45% of metastatic TCC and have been observed in other carcinomas as well.2 4 Unfortunately, markers that are specic for a single type of epithelium such as prostate-specic antigen for the prostate and thyroglobulin for the thyroid are rare. Recently, a new group of antigens (uroplakins [UPs]) specic for the urothelium

Uroplakins as a Marker for Metastatic TCC on FNA/Xu et al.

217

human tissue tested including tissues from the breast and salivary gland, oral and paranasal sinus mucosa, lung tissue, gastrointestinal tissue, and liver tissue.9 Immunohistochemical staining of the primary TCC with UP antibodies showed supercial membranous and microluminal staining patterns in 14 of 16 papillary noninvasive TCCs (88%) and 29 of 55 invasive TCCs (53%). One hundred six non-TCC carcinomas including breast, lung, ovarian, and gastrointestinal primary tumors consistently were negative.9

MATERIALS AND METHODS

Slides from 27 FNA cases of metastatic TCC and 52 cases of non-TCC carcinomas were collected from the cytopathology laboratory of the Hospital of the University of Pennsylvania from January 1, 1990 to December 31, 1999. FNAs were performed by either a clinician or a cytopathologist with immediate on-site evaluation of cellular adequacy, preliminary diagnosis, and subsequent nal report. The procedures were performed using a 23-gauge needle with 50% of the smears air-dried and stained by the Diff-Quik (EM Science, Gibbstown, NJ) method for immediate evaluation and the remaining smears wet-xed by immersion in 95% ethanol and stained with Papanicolaou stain. Representative 95% ethanol-xed, Papanicoloau-stained direct smears were selected for study from each case. Polyclonal antibody against UP II was used as the primary antibody after a 1:200 dilution. For the immunocytochemical staining of cytologic direct smears, the avidin-biotin-peroxidase complex method (Vector Laboratories, Burlingame, CA) was applied without destaining of the Papanicolaou smears.10 Cover slips rst were removed by xylene bath. These smears were rehydrated and endogenous peroxidase activity was blocked by incubation with 0.9% hydrogen peroxide and methanol for 15 minutes. After blocking, the smears were incubated with primary antibodies at a 1:200 dilution for 1 hour and Universal Secondary Antibody (Ventana, Tucson, AZ) for 45 minutes. The slides were washed and developed following the manufacturers instructions. Ninety-ve percent ethanolxed and Papanicolaou-stained direct smears of benign renal pelvic brush specimens (positive) and bronchial brush specimens (negative) were used as controls. The existing hematoxylin from the Papanicolaou stain served as the nuclear counterstain for all control and representative tumor slides evaluated in the current study. A total of 79 FNA cases were investigated in the current study, including 34 lymph nodes, 15 lung specimens, 7 liver specimens, 3 kidney specimens, 6 breast specimens, 13 subcutaneous and soft tissue specimens and 1 parotid gland specimen. The location of 27 metastatic TCC cases and patho-

FIGURE 1. Fine-needle aspiration specimen of metastatic transitional cell carcinoma containing cercariform cells characterized by tadpole-shaped malignant cells with eccentric nuclei and cytoplasmic features of both glandular and squamous cytoplasm (Papanicolaou-stained, 95% ethanol-xed direct smear).
have emerged; these antigens have the potential to identify metastatic TCC in patients presenting with an unknown primary tumor. UPs are a group of transmembrane proteins constituting the asymmetric unit membrane of urothelial umbrella cells. Four different UPs (UP Ia [27 kilodaltons (kD)], UP Ib [28 kD], UP II [15 kD], and UP III [47 kD]) have been demonstrated to be specic urothelial tissue markers.5,6 Studies utilizing in situ hybridization techniques have shown that normal urothelium expressed all four UP genes in parafn-embedded tissues.5 UP expression by TCC cell line studies in vitro was found to be related to their differentiated phenotype, with well-differentiated tumor RT4 cells expressing all four UPs, moderately differentiated VM-CUB-3 cells expressing three UPs, and poorly differentiated RT112 and HT1376 cells expressing UP Ib only.5 The observations of the TCC cell lines appeared to be well correlated with those of the parafn-embedded tissues in which although UP Ia and UP II were expressed in the supercial cells of well-differentiated, noninvasive papillary TCC, but only UP Ib was found to be expressed in seven of nine poorly differentiated invasive TCC and all ve metastatic TCC cases tested.5 Polyclonal antibodies to UP proteins were harvested after injecting rabbits with isolated asymmetric unit membranes or synthetic amino acid peptide.6 8 UP staining of normal formalin-xed, parafn-embedded tissue is conned to the supercial umbrella cells of the urothelium at all sites including the renal pelvis, ureter, urinary bladder, and prostatic urethra.9 In cases of cystitis cystica, luminal membrane staining was observed in some but not all of the microcystic structures.9 No UP staining was observed in the other

218

CANCER (CANCER CYTOPATHOLOGY) June 25, 2001 / Volume 93 / Number 3

TABLE 1 Case Data and Results of Uroplakin Immunohistochemical Staining

Cellular staining pattern No. of positive cases 11 4 3 1 6 2 SCC 0 0 0 0 0 1 SCC Supercial membrane/ microluminal

Location (TCC) or nature of lesion (non-TCC) TCC* Lymph nodes Lung Liver Kidney Subcutaneous and soft tissue Lung: 8 small cell, 6 AdenoCA and 5 SCC (8 P and 9 M) GI: 8 AdenoCA and 1 SCC (9 M) GYN: 5 AdenoCA and 1 SCC (6 M) Breast AdenoCA (6 P and 2M) Prostate AdenoCA (3 M) Renal cell CA (6 M) Urethral: 1 SCC (1 M)

No. of cases 12 4 3 1 7 19 9 6 8 3 6 1

Diffuse membranous (6 cases) (2 cases)

(4 cases)

Non-TCC

TCC: transitional cell carcinoma; : staining pattern observed in the majority of cellular material present; : staining pattern observed focally in the cellular material present; : no staining pattern observed in the cellular material present; AdenoCA: adenocarcinoma; SCC: squamous cell carcinoma; P: primary; M: metastatic; GI: gastrointestinal; GYN: gynecologic; CA: carcinoma.

logic data from the 52 non-TCC carcinomas (35 metastatic and 17 primary tumors), including 30 adenocarcinomas, 8 squamous cell carcinomas, 8 small cell carcinomas, and 6 renal cell carcinomas, are summarized in Table 1. The patients included 43 men and 36 women who ranged in age from 4396 years.

RESULTS
The results of the immunocytochemical staining of the tumors for UP antibody are summarized in Table 1. Of the 27 metastatic TCC cases, 25 (93%) stained positively. The pattern of staining was a consistent supercial membranous/microluminal pattern. Diffuse membranous staining was noted in a small proportion ( 10%) of cells in 50% of the positive metastatic TCC cases. None of the non-TCC carcinomas stained with the supercial membranous/microluminal pattern; however; 3 of the 52 cases (6%) exhibited rare tumor cells with diffuse membranous staining. These were keratinized squamous cell carcinomas (two from the lung and one from the urethra), but morphologically these cases were distinguished easily from metastatic TCC. In addition, in all three of these cases, only occasional tumor cells ( 1%) stained in a diffuse membranous pattern whereas the majority of the staining occurred in background keratin debris (Fig. 2). The most common observation in the metastatic TCC cases was supercial membranous staining, which was observed in all 25 positive cases (100%). The supercial membranous staining observed in metastatic TCC cases closely mimics the pattern pro-

FIGURE 2. Immunocytochemical staining of uroplakin in a primary squamous

cell carcinoma from a ne-needle aspiration specimen of the lung. Note the focal, nonspecic, diffuse cellular staining pattern of the tumor cells and the intense background staining of keratin debris (Uroplakin antigen-antibody reaction performed on Papanicolaou-stained, 95% ethanol-xed direct smear with diaminobenzidine chromogen).

duced in the positive controls of normal urothelial brush specimens, although UP staining generally was less intense and even (Fig. 3). The second most common pattern, microluminal staining, characterized by membranous staining of morphologic inconspicuous lumen within the neoplastic urothelium, was noted in 21 of 25 positive cases (84%) (Fig. 4). A third pattern of

Uroplakins as a Marker for Metastatic TCC on FNA/Xu et al.

219

FIGURE 4. Immunocytochemical stain for uroplakin in a metastatic transitional cell carcinoma from a ne-needle aspiration specimen from the liver. Note the microluminal staining pattern (uroplakin antigen-antibody reaction performed on Papanicolaou-stained, 95% ethanol-xed direct smear with diaminobenzidine chromogen).

FIGURE 5. Immunocytochemical stain of a metastatic transitional cell carcinoma from a retroperitoneal lymph node. This gure demonstrates diffuse membranous staining (uroplakin antigen-antibody reaction performed on Papanicolaou-stained, 95% ethanol-xed direct smear with diaminobenzidine chromogen).

diffuse membranous staining was observed in a few cells in 12 of 25 positive cases (48%) (Fig. 5). The observation of one of the three specic staining patterns was considered a positive nding in the current study.

FIGURE 3. (A) Immunocytochemical staining for uroplakin in normal urothelium

from a bladder brush specimen. Notice the intense continuous luminal membrane staining of the umbrella cells. (B) Immunocytochemical staining for uroplakin in metastatic transitional cell carcinoma from a ne-needle aspiration specimen of a pelvic lymph node demonstrating a supercial membrane pattern that was less intense and less continuous than that seen in Figure 1A (uroplakin antigen-antibody reaction performed on Papanicolaou-stained, 95% ethanol-xed direct smear with diaminobenzidine chromogen).

DISCUSSION
UPs are integral membrane components that are synthesized by terminally differentiated urothelial cells.6,7,9 Their functions include a permeability barrier and, through their binding to the cytoskeleton, they may stabilize the urothelial luminal surface and prevent rupturing during urinary tract distention.11,12

220

CANCER (CANCER CYTOPATHOLOGY) June 25, 2001 / Volume 93 / Number 3

To our knowledge UPs are expressed only in the urothelium and have not been detected in any other tissues examined to date.9,12 Tissues such as the stomach, which, like the bladder, can expand tremendously, do not appear to express UPs. Loss of differentiation during the development of urothelial malignancy may be reected in the loss of UP expression.13 This may account for the less intense and less continuous supercial membrane staining relative to the benign urothelial controls noted in the current study. Ogawa et al. reported the patterns of UP expression in various degrees of chemically induced hyperplasia and carcinoma in the rat bladder.14 In hyperplasia, the expression of UPs was observed in the intermediate cells as well as in the normal location on the luminal surface but the staining patterns remained orderly. In carcinomas, UP expression was absent from the luminal surface; it became disorganized with only a small amount of focal staining in the intermediate cells. Recently, the detection of UP-positive cells in the blood circulation was proposed to be a strong indication of the hematogenous dissemination of urothelial tumors among patients with TCC.15 However, it is necessary to remember that UPs are not markers of the biologic behavior of the tumor or the patients prognosis. UPs are detectable in the majority of TCC cases regardless of the grade and stage of disease. It has been shown that a signicant proportion (77 88%) of primary urothelial-derived tumors express UPs, although only 66% of metastatic TCCs are UP positive when examined using parafn-embedded tissue.5,9,13 The current study demonstrates the increased sensitivity (93%) of immunocytochemical stains for UP performed on 95% ethanol-xed, Papanicolaou-stained direct smears of metastatic TCCs, a gure that is considerably better than that reported previously utilizing immunohistochemical stains for UP on formalin-xed, parafn-embedded tissues from metastatic TCC (66%).9 The increased sensitivity of immunocytochemical stained, ethanol-xed direct smears relative to immunohistochemistry performed on parafn-embedded tissue is not surprising and has been reported with other antigens such as cytokeratins.10,16 The specicity of the 95% ethanol-xed direct smear is comparable to that for formalin-xed, parafn-embedded tissue, especially if one observes the supercial membranous/microluminal pattern. In addition, relative to searching for cercariform cells (previously reported to be present in only 45% of FNA specimens diagnosed with metastatic TCC), the 93% detection rate for UPs noted in the current study may improve our ability to recognize metastatic malignancies of urothelial orgin.4 The majority of the non-TCC FNA tissues in the

current study lacked any detectable UP immunoreactivity. Particularly interesting are the negative reactions observed in the metastatic renal cell and prostate carcinoma specimens. The three cases of squamous cell carcinoma with extensive mixed inammatory inltrates, contained rare cells with diffuse membranous staining and no supercial membranous or microluminal staining. These ndings may be considered false-positive results. It is unclear whether this is due to inammatory inltration and insufcient quench of the enzymatic activity or the presence of related, non-UP epitopes. Similar phenomena also were observed when using formalin-xed, parafn-embedded tissues.9 UPs are a sensitive and specic marker for a tumor of urothelial origin when applied to ethanol-xed direct smear FNA material, although it must be acknowledged that the presence of UP may be very focal in certain cases and that some poorly differentiated TCCs may not express UP.17 Therefore, a negative immunocytochemical result does not necessarily exclude the possibility of a poorly differentiated TCC. It has been shown that UP gene expression by in situ hybridization was detected in some poorly differentiated invasive TCCs and metastatic TCCs.5 It is possible that the combination of immunocytochemical staining and in situ hybridization of UP genes may improve our ability to determine the urothelial origin of poorly differentiated tumors further.

REFERENCES
1. Lamm DL, Grifth JG. Intravesical chemotherapy: does it affect the natural history of supercial bladder cancer? Semin Urol 1992;10:39 42. Powers CN, Elbadawi A. Cercariform cells: a clue to the cytodiagnosis of transitional cell origin of metastatic neoplasms? Diagn Cytopathol 1995;13:1521. Renshaw AA, Madge R. Cercariform cells for helping distinguish transitional cell carcinoma in ne needle aspirates. Acta Cytol 1997;41:999 1007. Hida HA, Gupta PK. Cercariform cells: are they specic for transitional cell carcinoma? Cancer (Cancer Cytopathol) 1998;87:69 74. Lobban ED, Smith BA, Hall GD, Harnden P, Roberts P, Selby PJ, et al. Uroplakin gene expression by normal and neoplastic human tissue. Am J Pathol 1998;153:1957 67. Wu XR, Lin JH. Mammalian uroplakins: a group of highly conserved urothelial differentiation-related membrane proteins. J Biol Chem 1994;269:13716 24. Wu XR, Manabe M, Yu J, Sun TT. Large scale purication and immunolocalization of bovine uroplakins I, II and III: molecular markers of urothelial differentiation. J Biol Chem 1990;265:19170 9. Yu J, Manabe M, Wu XR, Xu C, Surya B, Sun TT. Uroplakin I: a 27-kD protein associated with the asymmetric unit membrane of mammalian urothelium. J Cell Biol 1990;111:120716.

2.

3.

4.

5.

6.

7.

8.

Uroplakins as a Marker for Metastatic TCC on FNA/Xu et al.

9. Moll R, Wu XP, Lin JH, Sun TT. Uroplakins specic membrane proteins of urothelial umbrella cells, as histologic markers of metastatic transitional cell carcinoma. Am J Pathol 1995;147:138397. Gupta PK, Myers JD, Baylin SB, Mushine JL, Cuttitta F, Gazdar AF. Improved antigen detection in ethanolxed cytologic specimens: a modied avidin-biotin-peroxidase complex (ABC) method. Diagn Cytopathol 1985; 1:133 6. Minsky BD, Chlapowski FJ. Morphometric analysis of the translocation of luminal membrane between cytoplasm and cell surface of transitional epithelial cells during the expansion-contraction cycles of mammalian urinary bladder. J Cell Biol 1978;77:6859. Sun TT, Liang FX, Wu XR. Uroplakins as markers of urothelial differentiation. Adv Exp Med Biol 1999;462:718. Wu RL, Osman I, We XR, Lu ML, Zhang ZF, Liang FX, et al. Uroplakin II gene is expressed in transitional cell carcinoma

221

14.

10.

15.

11.

16.

12. 13.

17.

but not in bilharzial bladder squamous cell carcinoma: alternative pathways of bladder epithelial differentiation of tumor formation. Cancer Res 1998;58:12917. Ogawa K, Sun TT, Cohen SM. Analysis of differentiationassociated proteins in rat bladder carcinogenesis. Carcinogenesis 1996;17:9615. Li SM, Zhang ZT, Chan S, Mclenan O, Dixon C, Samir T, et al. Detection of circulating uroplakin-positive cells in patients with transitional cell carcinoma of the bladder. J Urol 1999;162:9315. Taylor CR. Immunohistology of the skin: melanoma, intermediate laments. In: Taylor CR, editor. Immunomicroscopy: a diagnostic tool for the surgical pathologist. Philadelphia: W.B. Saunders Company, 1986:290 1. Finch JL, Miller J, Aspinall JO, Cowled P. Cloning of the human uroplakin 1B cDNA and analysis of its expression in urothelial-tumor cell lines and bladder carcinoma tissue. Int J Cancer 1999;80:533 8.