Craniopharyngiomas are epithelial neoplasms thought

Differential Expression of Cytokeratins 8 and 20 Distinguishes Craniopharyngioma From Rathke Cleft Cyst Wei Xin, MD, PhD; Mark A. Rubin, MD; Paul E. McKeever, MD, PhD Background. Craniopharyngiomas are epithelial neoplasms usually located in the sellar and suprasellar regions. Distinguishing craniopharyngioma from Rathke cleft cyst is sometimes difficult, and the distinction is clinically significant because Rathke cleft cysts have a better prognosis than craniopharyngiomas. Design. We retrieved 10 cases with a primary diagnosis of craniopharyngioma and 5 cases with a diagnosis of Rathke cleft cyst for analysis. Five cases of normal pars intermedia of pituitary glands from autopsy served as controls. We evaluated the expression patterns of a broad range of low to intermediate molecular weight cytokeratins (CK7, CK8, CK10, CK17, CK18, CK19, and CK20) and high molecular weight cytokeratins (K903: a combination of CK1, CK5, CK10, and CK14; and CK5/6) in these cases. Results. Craniopharyngiomas had a cytokeratin expression pattern distinct from that of Rathke cleft cysts and pituitary gland pars intermedia: craniopharyngiomas did not express cytokeratins 8 and 20, whereas Rathke cleft cysts and pars intermedia of pituitary glands both expressed cytokeratins 8 and 20. Conclusion. The differential expression of cytokeratins distinguishes between craniopharyngioma and Rathke cleft cyst, and this difference could be useful for identifying craniopharyngioma in difficult cases in which only a small biopsy is available. The different cytokeratin profiles of craniopharyngioma and Rathke cleft cyst suggest that these lesions do not come from the same origin, or that they come from a different developmental stage of the pouch epithelium. (Arch Pathol Lab Med. 2002;126:1174 1178) Craniopharyngiomas are epithelial neoplasms thought to arise from the remnants of the Rathke pouch. They are usually located in the sellar and suprasellar regions. Craniopharyngiomas account for 1% to 2% of all intracranial neoplasms, but they are the most common tumor of childhood and adolescence. 1 3 They are histologically benign neoplasms with an infiltrative growth pattern. They are often difficult to excise surgically, and they have a 10% to 40% recurrence rate. Craniopharyngiomas compress surrounding structures to cause visual disturbance, hydrocephalus, and endocrine abnormalities. 4 Craniopharyngiomas are tumors with a variable, occasionally predominant, cystic component that are subclassified into 2 clinicopathologic entities: the adamantinomatous and the papillary craniopharyngiomas. 5 Rathke cleft cysts are also thought to originate from the remnants of the Rathke pouch. Histologically, Rathke cleft cysts are lined by a ciliated cuboidal or columnar epithelium with common squamous metaplasia. Rathke cleft Accepted for publication May 28, 2002. From the Departments of Pathology (Drs Xin, Rubin, and McKeever) and Urology (Dr Rubin), and the Comprehensive Cancer Center, University of Michigan Medical Center, Ann Arbor. Presented in part at the 2001 annual meeting of College of American Pathologists/American Society of Clinical Pathologists, Philadelphia, Pa, October 19 22, 2001. Reprints: Wei Xin, MD, PhD, Department of Pathology, University of Michigan Medical Center, 1500 E Medical Center Dr, Ann Arbor, MI 48109 (e-mail: weixin@umich.edu). cysts are common. 6 Occasionally, these cysts enlarge to become clinically detectable and cause compressive effects like craniopharyngiomas. 7 9 The prognosis of Rathke cleft cysts with surgery is excellent, with a very low recurrence rate. 1,7 In many instances, the differentiation between craniopharyngioma and Rathke cleft cyst is not difficult when substantial tissue is provided. However, in other cases, a specimen may show only small clusters of squamous or columnar neoplastic cells and no nodule of craniopharyngioma cells or unambiguous lining cells of Rathke cleft cyst. In such cases, it can be difficult to differentiate these 2 entities. It has been accepted that cell type specificity of intermediate filaments is conserved during cell transformation and tumor development. Cytokeratins (CKs), a complex family of many different polypeptides, are expressed in epithelia and their related neoplasms. A particular epithelium can be characterized by its specific pattern among the highly diverse expression patterns of cytokeratins. 10,11 Recent studies have emphasized CK7 and CK20 on phenotyping tumors. 12 14 In previous reports, lesions of both craniopharyngiomas and Rathke cleft cysts have been found to be immunoreactive for cytokeratins. 15,16 In this study, we investigated the diverse cytokeratin expression patterns in surgically excised craniopharyngiomas and Rathke cleft cysts, as well as in pituitary gland pars intermedia from necropsy cases. We used a broad range of low to intermediate molecular weight cytokeratins (CK7, CK8, CK10, CK17, CK18, CK19, and CK20) 1174 Arch Pathol Lab Med Vol 126, October 2002 CKs in Craniopharyngioma and Rathke Cleft Cyst Xin et al

Table 1. Monoclonal Antibodies and Methods for Immunohistochemistry CK* Titer Clone Antigen Retrieval Positive Control Tissue Automated Stainer K903 CK5/6 CK7 CK8 CK10 CK17 CK18 CK19 CK20 1:12.5 1:25 34 E12 D5-16B4 OV-TL 12/30 35BH11 DE-K10 E3 DC10 BA17 Ks20.8 * CK indicates cytokeratin. Treated with microwave. Recommended and provided by Dako Corporation. Tris buffer, ph 9.0 Enzyme II None Citrate buffer, ph 6.0 None Enzyme I Prostate Tonsil Lung adenocarcinoma Skin Skin Breast ductal carcinoma Prostate Breast ductal carcinoma Colon adenocarcinoma Ventana 320 Ventana 320 Table 2. Craniopharyngioma (n 10) Rathke cleft cyst (n 5) Pituitary gland pars intermedia (n 5) Cytokeratin (CK) Expression in Craniopharyngioma, Rathke Cleft Cyst, and Pituitary Gland Pars Intermedia* K903 CK5/6 CK7 CK8 CK10 CK17 CK18 CK19 CK20 80 (2 3 ) 10 (1 ) 100 (2 ) 10 (1 ) 20 (1 ) 20 (1 ) 100 (2 ) * Values are percentages of tumor cells stained: 5% of cells stained ( ), 5% to 25% (1 ), 25% to 50% (2 ), and 50% (3 ). 60 (1 2 ) and high molecular weight cytokeratins (K903: a combination of CK1, CK5, CK10, and CK14; and CK5/6). MATERIALS AND METHODS Case Selection From our files, we identified 10 cases of craniopharyngioma (female-male ratio, 7:3) and 5 cases of Rathke cleft cyst (femalemale ratio, 2:3) from the years 1988 1998. All of the diagnoses were reconfirmed, and only cases of primary diagnosis were included in the study. We also selected 5 normal pituitary gland pars intermedia specimens from autopsy cases as controls. The craniopharyngioma group consisted of 5 female children (aged 3 12 years), 1 male adolescent (aged 18 years), and 2 male and 2 female adults (aged 39 62 years). The Rathke cleft cyst group consisted of 1 male adolescent (aged 17 years) and 2 male and 2 female adults (aged 28 58 years). The control necropsy group consisted of 3 male adults (aged 54 70 years) and 2 female adults (aged 55 69 years). Immunohistochemistry We purchased commercially available antibodies K903 (CK1, CK5, CK10, and CK14) (clone 34 E12), CK7 (clone OV-TL 12/ 30), CK8 (clone 35BH11), CK10 (clone DE-K10), CK17 (clone E3), CK18 (clone DC10), CK19 (clone BA17), and CK20 (clone Ks20.8) from Dako Corporation (Carpinteria, Calif). We purchased CK5/ 6 (clone D5-16B4) from Chemicon International Inc (Temecula, Calif). All of the sections used in the study were from formalinfixed, paraffin-embedded tissue. The sections were deparaffinized and rehydrated in graded alcohols. The titers, antigen retrieval, and automated stainers used are summarized in Table 1. For negative controls, we replaced primary antibodies with nonimmunized mouse serum. We assessed the immunoreactivity of tissues for cytokeratins under a light microscope (model U-DO, Olympus Optical Corporation, Japan). We graded the extent of immunoreactivity based on the percentage of tumor cells that stained for the marker, as follows: less than 5% of the cells stained ( ), 5% to 25% (1 ), 25% to 50% (2 ), and greater than 50% (3 ). RESULTS Immunohistophenotype of Craniopharyngioma, Rathke Cleft Cyst, and Pituitary Gland Pars Intermedia The patterns of expression of different cytokeratins in craniopharyngiomas, Rathke cleft cyst, and pituitary gland pars intermedia are summarized in Table 2. All 3 entities were positive for high molecular weight cytokeratins (K903 and CK5/6) and for low to intermediate molecular weight cytokeratins CK7, CK17, and CK19. All 3 tissues were negative for intermediate molecular weight CK10 and seldom showed expression of CK18 (10% to 20% of specimens). However, the reactivity of craniopharyngioma for CK8 and CK20 was distinct from that of Rathke cleft cyst and pituitary gland pars intermedia. Neither 8 adamantinomatous nor 2 papillary craniopharyngiomas had detectable CK20 expression, and only 1 craniopharyngioma expressed CK8 (Figure 1). In contrast, all 5 Rathke cleft cysts including samples with squamous metaplasia had immunoreactivity to these 2 cytokeratins (Figure 2). Therefore, immunoreactivity for CK8 and CK20 distinguished Rathke cleft cysts from craniopharyngiomas. Pituitary gland pars intermedia (Figure 3) had a pattern of cytokeratin expression similar to that of Rathke cleft cyst (Figure 2) but different from that of craniopharyngioma (Figures 1). The adjacent cortical pituitary glands (Figure 2, B and C; Figure 3, B and C) were focally CK8 positive and diffusely and weakly CK20 positive. This pattern represents similarity of the gland to Rathke cleft cyst and pituitary pars intermedia rather than to craniopharyngioma. COMMENT Craniopharyngioma is a cystic epithelial tumor of the sellar region thought to arise from the Rathke pouch. Two variants have been described: adamantinomatous and Arch Pathol Lab Med Vol 126, October 2002 CKs in Craniopharyngioma and Rathke Cleft Cyst Xin et al 1175

Figure 1. Hematoxylin-eosin stain (A) and immunohistochemical stains for cytokeratin 8 (B) and cytokeratin 20 (C) in craniopharyngioma (adamantinomatous-type) under microscopic high-power magnification (original magnification 200). Figure 2. Hematoxylin-eosin stain (A) and immunohistochemical stains for cytokeratin 8 (B) and cytokeratin 20 (C) in Rathke cleft cyst under microscopic high-power magnification (original magnification 200). Figure 3. Hematoxylin-eosin stain (A) and immunohistochemical stains for cytokeratin 8 (B) and cytokeratin 20 (C) in pituitary gland pars intermedia under microscopic high-power magnification (original magnification 200). papillary. Adamantinomatous craniopharyngiomas have broad strands, cords, and bridges of a multistratified squamous epithelium with peripheral palisading of nuclei. Nodules of compact wet keratin and dystrophic calcification are diagnostic features. The cysts are sometimes lined by a simple flattened epithelium. Papillary craniopharyngiomas consist of sheets of squamous epithelium that separate to form pseudopapillae. Nuclear palisading, wet keratin, and dystrophic calcification are not typically present in this variant. 5 One report has stated that brain invasion is more common for adamantinomatous craniopharyngiomas than for papillary craniopharyngiomas, and that the papillary variants have a better prognosis, 17 whereas another report failed to support this conclusion. 18 In our study, 2 cases were papillary craniopharyngioma, and these cases showed a cytokeratin immunophenotype similar to that of adamantinomatous tumors, which are negative for CK8 and CK20. Rathke cleft cysts are derivatives of the Rathke pouch. The typical cells lining the cyst are cuboidal or columnar and ciliated cells admixed with goblet cells, but these can undergo metaplasia in the course of growth. 2 Squamous metaplasia is common in Rathke cleft cyst, and 2 cases (40%) in our study had unequivocal metaplastic squamous cells. Nevertheless, we found no discrepancies in immunophenotype between the squamous metaplasia cells and the columnar cells; all of these cells were CK8 and CK20 positive. The histologic appearances of craniopharyngioma and Rathke cleft cyst are quite distinctive in many instances, such as when a nodule of craniopharyngioma cells is evident, or when an unambiguous cyst lined with pseudostratified columnar cells intermingled with goblet cells is present. However, some small biopsy specimens of cyst wall lack this distinctive architecture. Faced with this problem, the pathologist need not ignore alternatives for confirming a diagnosis. It is important to differentiate between these 2 entities since they have different treatments and different prognoses. Craniopharyngioma is histologically benign but behaves aggressively, recurs frequently, and often requires radiation therapy. A local surgical excision nearly always cures Rathke cleft cyst. The Rathke pouch closes in an early stage of normal embryonic development, but its apical extremity persists into postnatal life as a cleft in the pars intermedia of the pituitary gland. The cleft is lined by a simple cuboidal or 1176 Arch Pathol Lab Med Vol 126, October 2002 CKs in Craniopharyngioma and Rathke Cleft Cyst Xin et al

columnar, often ciliated epithelium. Mucous goblet cells and squamous metaplasia are often present. 2 Traditionally, both Rathke cleft cysts and craniopharyngiomas were thought to arise from the Rathke pouch epithelium homologous to the pars intermedia of the pituitary gland. 1,2 The reason why these 2 entities presumed to have the same origin behave so differently is not clear and has never been addressed. Are these 2 entities truly embryonically similar? To address this question, we studied their cytokeratin expression patterns and compared these patterns to those of the pars intermedia of the pituitary gland. Cytokeratin expression patterns identify different types of epithelial cells. Among 20 currently identified human cytokeratins, 5 (CK7, CK8, CK18, CK19, and CK20) are expressed mainly in simple epithelia. CK7, CK8, and CK18 are found in trachea, urothelium of bladder, and several complex glands. CK20 is found mainly in intestinal epithelium, gastric foveolar epithelium, urothelium, and Merkel cells. 9,10 These epithelial cytokeratin patterns are highly conserved during neoplastic transformation. Rathke cleft cysts with or without the squamous metaplasia and the pars intermedia of the pituitary gland expressed simple epithelium cytokeratins CK7, CK8, CK19, and CK20, and most of them did not express CK18. Their similar cytokeratin expression patterns suggest that Rathke cleft cysts arise from the Rathke pouch remnants. In contrast, craniopharyngiomas expressed CK7 and CK19 only, and lacked CK8 and CK20 expressions. One CK8- positive craniopharyngioma was of the adamantinomatous subtype, and its immunoreactivity was barely 1. This finding raises an intriguing question about the true origin of craniopharyngiomas. Craniopharyngiomas and Rathke cleft cysts might arise from different origins entirely, or both may arise from Rathke pouch epithelium but at different stage of embryonic development. Distinguishing these 2 entities is difficult when only a small number of atypical cells are present. However, in this context, positivity for CK8 and CK20 rules out craniopharyngioma. Clearly, negativity for CK8 and CK20 might suggest the diagnosis of craniopharyngioma. However, a diagnosis based on negativity is more tenuous because of factors that mimic negativity, including loss of antigen integrity and technical problems with the staining procedure. Therefore, a criterion of positivity of at least 5% cells stained was preferred to boost the sensitivity to rule out Rathke cleft cyst. Immunohistochemical staining revealed CK-20 positive cases both in craniopharyngioma and Rathke cleft cyst in which staining was at least 2, whereas only 1 case of pituitary gland pars intermedia (20%) had weak (1 ) CK20 positivity. The reason for the weak expression of CK20 in this case was not clear, but some autopsy specimens (pituitary gland pars intermedia) showed weak CK20 staining. Some specimens of sellar or suprasellar cyst walls are received with no solid tumor, but a lining clearly too large to be normal pars intermedia. This is the context in which CK8 and CK20 staining are potentially useful. Positivity for these markers would favor Rathke cleft cyst. Negativity for these markers would be indeterminate, but if staining for both was negative, the unlikely occurrence of 2 falsenegative stains would favor a diagnosis of craniopharyngioma, especially if the tissue was positive for the K903, CK17, and CK19 markers. Should a specific positive marker for craniopharyngioma be found in the future, it would be a useful addition to a staining panel to differentiate between these entities. The pituitary glands were focally CK8 positive and diffusely CK20 positive (Figure 2, B and C; Figure 3, B and C). This pattern indicates similarity of the gland to Rathke cleft cyst and pituitary pars intermedia rather than to craniopharyngioma. Although cytokeratin profiles will distinguish craniopharyngioma from pituitary gland, they will not distinguish Rathke cleft cyst from pituitary gland. It is clear that cytokeratin profiling will not solve every problem the pathologist encounters with small or fragmented specimens of these entities. These profiles should be considered along with other ancillary evidence in cases in which questions remain after hematoxylin-eosin staining. We found that adamantinomatous and papillary variants of craniopharyngiomas have the same cytokeratin profiles, and we found no evidence that they are histogenetically different. We used monoclonal antibody clone 35BH11, which is reported to be a specific anti-ck8 antibody (Table 1), and found that only 1 adamantinomatous craniopharyngioma expressed CK8. In contrast, a recent study found that all adamantinomatous craniopharyngiomas were CK8 positive, and 40% of papillary craniopharyngiomas were CK8 positive. 19 Also in that study, ciliated cuboidal cells and squamous metaplasia cells were CK8 positive and negative, respectively. Clone CAM5.2 was used in that study as the anti-ck8 antibody. CAM5.2 is a combined subtype antibody reacting with both CK8 and CK7 that was previously reported to react with CK8, CK18, and CK19. 20 It is possible that CAM5.2 immunoreactivity for CK7 influenced the staining results in craniopharyngiomas. Other differences between the study by Tateyama et al 19 and our study require further exploration. In conclusion, differential expression of CK8 and CK20 distinguishes between craniopharyngioma and Rathke cleft cyst. This difference could be useful for making a diagnosis in cases that are difficult to identify with less decisive methods. Craniopharyngioma and Rathke cleft cyst might not arise from the same origin, or they might arise from a same origin but from a different developmental stage of the pouch epithelium. The authors would like to thank Michelle Leblanc from the Comprehensive Cancer Center Immunohistochemistry core for her technical assistance. This work was supported in part by grants CA R01-68545 and 3P30 CA 46592 from the United States Public Health Service. References 1. Sternberg SS, Antonioli DA, Carter D, Mills SE, Oberman HA, eds. Diagnostic Surgical Pathology. 3rd ed. Baltimore, Md: Williams & Wilkins; 1999:516 529. 2. Russel DS, Rubinstein LJ. Pathology of Tumours of the Nervous System. 5th ed. Baltimore, Md: Williams & Wilkins; 1989:702 704. 3. Laws ER Jr. Transsphenoidal microsurgery and management of craniopharyngioma. J Neurosurg. 1980;52:661 666. 4. Petito CK, DeGirolami U, Earle KM. Craniopharyngiomas: a clinical and pathological review. Cancer. 1976;37:1944 1952. 5. Janzer RC, Burger PC, Giangaspero F, Paulus W, eds. Craniopharyngioma. In: Kleihues P, ed. World Health Organization Classification of Tumors: Pathology and Genetics of Tumors of the Nervous System. Lyon, France: IARC Press; 2000: 244 246. 6. Keyaki A, Hirano A, Llena JF. Asymptomatic and symptomatic Rathke s cleft cysts: histological study of 45 cases. Neurol Med Chir (Tokyo). 1989:29:88 93. 7. Asa SL. Tumors of the Pituitary Gland. Washington DC: Armed Forces Institute of Pathology; 1988:167 176, 197 198. 8. Steinberg GK, Koenig GH, Golden JB. Symptomatic Rathke s cleft cysts: report of two cases. J Neurosurg. 1982;56:290 295. 9. Voelker JL, Campell RL, Muller J. Clinical, radiographic, and pathologic features of symptomatic Rathke s cleft cysts. J Neurosurg. 1991;74:535 544. Arch Pathol Lab Med Vol 126, October 2002 CKs in Craniopharyngioma and Rathke Cleft Cyst Xin et al 1177

10. Moll R, Franke WW, Schiller DL. The catalog of human cytokeratins: patterns of expression in normal epithelia, tumor and cultured cells. Cell. 1982;31: 11 24. 11. Moll R, Lowe A, Laufer J, et al. Cytokeratin 20 in human carcinomas. Am J Pathol. 1992;140:427 447. 12. Chu P, Wu E, Weiss L. Cytokeratin 7 and cytokeratin 20 expression in epithelial neoplasms: a survey of 435 cases. Mod Pathol. 2000;13:962 972. 13. Wang NP, Zee S, Zarbo RJ, et al. Coordinate expression of cytokeratins 7 and 20 defines unique subsets of carcinomas. Appl Immunohistochem. 1995;3: 99 107. 14. Maeda T, Kajiyama K, Adachi E, et al. The expression of cytokeratins 7, 19 and 20 in primary and metastatic carcinomas of the liver. Mod Pathol. 1996;9: 901 902. 15. Uemastsu Y, Norihiko K, Hayashi S, et al. Cytokeratin immunohistochemical study of epithelial cysts in the central nervous system: with special reference to origins of colloid cyst of the third ventricle and Rathke s cleft cyst in sella. Brain Tumor Pathol. 1993;10:43 52. 16. Kubo O, Tasuhiko Y, Takakura K, et al. Immunohistochemical study of craniopharyngiomas. Brain Tumor Pathol. 1993;10:131 134. 17. Adamson TE, Wiestler OD, Kleihues P, Yasargil MG. Correlation of clinical and pathological features in surgically treated craniopharyngiomas. J Neurosurg. 1990;73:12 17. 18. Crotty TB, Scheithauer BW, Young WF, et al. Papillary craniopharyngioma: a clinicopathological study of 48 cases. J Neurosurg. 1995;83:206 214. 19. Tateyama H, Tad T, Okabe M, et al. Different keratin profiles in craniopharyngioma subtypes and ameloblastomas. Pathol Res Pract. 2001;197:735 742. 20. Makin C, Bobrow L, Bodmer W. Monoclonal antibody to cytokeratin for use in routine histopathology. J Clin Pathol. 1984;37:975 983. 1178 Arch Pathol Lab Med Vol 126, October 2002 CKs in Craniopharyngioma and Rathke Cleft Cyst Xin et al