doi: 10.1111/jop.12234
J Oral Pathol Med © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd wileyonlinelibrary.com/journal/jop
Ghost cells in pilomatrixoma, craniopharyngioma, and calcifying cystic odontogenic tumor: histological, immunohistochemical, and ultrastructural study Alicia Rumayor1, Rom an Carlos2, Hern an Molina Kirsch3, Bruno A. Benevenuto de Andrade4, Mario J. 4 1 Roma~ nach , Oslei Paes de Almeida 1
Oral Pathology Section, Department of Oral Diagnosis, Piracicaba Dental School, University of Campinas/UNICAMP, Piracicaba, Brazil; 2Pathology Division, Centro Clınico de Cabeza y Cuello/Hospital Herrera Llerandi, Guatemala City, Guatemala; 3Pathology Laboratory, Guatemala City, Guatemala; 4Oral Pathology Section, Department of Oral Diagnosis and Pathology, School of Dentistry, Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, Brazil
BACKGROUND: Pilomatrixoma, craniopharyngioma, and calcifying cystic odontogenic tumor are the main entities presenting ghost cells as an important histological feature, in spite their quite different clinical presentation; it seems that they share a common pathway in the formation of these cells. The aim of this study is to examine and compare the characteristics of ghost and other cells that form these lesions. METHODS: Forty-three cases including 21 pilomatrixomas, 14 craniopharyngiomas, and eight calcifying cystic odontogenic tumors were evaluated by immunohistochemistry for cytokeratins, CD138, b-catenin, D2-40, Glut-1, FAS, CD10 and also by scanning electron microscopy. RESULTS: The CKs, CD138, b-catenin, Glut-1, FAS, and CD10 were more often expressed by transitional cells of craniopharyngioma and calcifying cystic odontogenic tumor, compared with pilomatrixoma. Basaloid cells of pilomatrixoma showed strong positivity for CD138 and CD10. Differences on expression pattern were identified in transitional and basal cells, as ghost cells were negative for most antibodies used, except by low expression for cytokeratins. By scanning electron microscopy, the morphology of ghost cells were similar in their fibrillar cytoplasm, but their pattern varied from sheets in pilomatrixoma to small clusters in craniopharyngioma and calcifying cystic odontogenic tumor. CONCLUSIONS: Mechanisms involved in formation of ghost cells are unknown, but probably they follow different pathways as protein expression in the basal/
Correspondence: Alicia Rumayor, DDS, MSc, Oral Pathology, Piracicaba Dental School, University of Campinas (UNICAMP), Av. Limeira 901, P.O. Box 52, 13414-903, Piracicaba, S~ao Paulo, Brazil. Tel: +551921065315, Fax: +551921065218, E-mail: [email protected] Accepted for publication June 17, 2014
transitional cells was not uniform in the three tumors studied. J Oral Pathol Med (2014) Keywords: calcifying cystic odontogenic tumor; craniopharyngioma; ghost cells; immunohistochemistry; pilomatrixoma
Introduction Ghost cells (GC) are characterized by their abundant eosinophilic cytoplasm, a central empty round space which corresponds to the area previously occupied by the nuclei and a well-defined outline. They are typically observed in calcifying cystic odontogenic tumor (CCOT), craniopharyngioma (CP), pilomatrixoma (PM) and eventually can also be found in ameloblastoma, odontoma and ameloblastic fibro-odontoma (1–3). Although the mechanisms involved in GC formation are unknown; most authors consider it an unusual or aberrant form of cellular keratinization, called ‘ghost cell keratinization’. By immunohistochemistry, CKs in GC vary in type and intensity. Besides, enamel-related proteins have been described in GC of CCOT, and human hair proteins in PM, CP and CCOT (4–7). Ultrastructural studies of PM have shown bundles of fibrils within the cytoplasm of GC, interpreted as tonofilaments, probably related to keratins. A similar fibrillar pattern was also described in GC of CCOT. The loss of the nucleus in the cytoplasm of GC generates an empty round central space identified as an electron-lucent area by transmission electron microscopy (TEM) (8–11). The objective of this study is to describe and compare the histopathological, immunohistochemical, and ultrastructural features of GC and its precursors in PM, CP, and CCOT.
Ghost cells study in three tumors Rumayor et al.
2
Table 1 Antibodies used for the immunohistochemical analysis
Antibody
Clone
Source
Dilution
AE1/AE3
AE1/AE3
Dako
1:500
CK8 CK14 CK19 CD138 b-catenin
TS1 LL002 RCK 108 My 15 17 C 2
Novocastra Novocastra Dako Dako Novocastra
1:500 1:200 1:200 1:100 1:200
D2-40 Glut-1 FAS CD10
D2-40 Polyclonal Polyclonal 56C6
Dako Bio Systems Atlas Novocastra
1:100 1:100 1:200 1:100
Antigen specificity Cocktail/Epithelial cell Epithelial cell Epithelial cell Epithelial cell Intercellular junction Intercellular junction Wnt signaling pathway Basal epithelial cells Glucose transporter Fatty acid synthesis Cell surface metalloproteinase
Materials and methods Forty-three cases including 21 PM, 14 CP, and eight CCOT were used in this study. Histological 3-lm-thick sections of paraffin-embedded tissue were stained with H&E and processed for conventional immunohistochemistry and scanning electron microscopy (SEM). For immunohistochemistry, sections were dewaxed with xylene and then hydrated in ethanol. After antigen retrieval, endogenous peroxidase activity was blocked using five baths of 10% hydrogen peroxide, 5 min each. Antigen retrieval was performed in a pressure cooker with citrate buffer (pH
6.0), except for FAS and CD10 in which EDTA/Tris (pH 9.0) was used. After being washed in phosphate-buffered saline (pH 7.4), slides were incubated overnight with primary antibodies (Table 1). All slides were subsequently exposed to avidin–biotin complex and peroxidase reagents (LSAB kit; Dako Cytomation, Glostrup, Denmark) and diaminobenzidine tetrahydrochloride (Sigma, St. Louis, MO, USA), and subsequently counterstained with Carazzi hematoxylin. Adequate positive control sections were used for each antibody, and the negative control was obtained by omitting the primary specific antibody. A descriptive analysis of histopathological and immunohistochemical findings was performed for all the markers, considering GC and its precursors. For SEM, 7-lm-thick sections of paraffin-embedded tissue were dewaxed with xylene, dried with ethanol, then mounted on a metallic stub, and coated with a thin layer of gold. Images were taken at a voltage of 15 kV with magnifications from 400 to 2500 times. The current study was approved by the Ethical Committee of Piracicaba Dental School, State University of Campinas.
Results Histopathologically, 90% of PM presented a solid pattern, with only two cases showing cystic features. Ten and nine cases of solid PM were classified as early and late regressive stage, respectively. Basaloid cells showed intensely basophilic nuclei, scanty cytoplasm, and indistinct cellular borders, a distinct transitional region between basaloid and GC was evident, formed by cells with a pyknotic nuclei
A
B
C
D
Figure 1 Histopathological aspects of PM, CP, and CCOT. (A) PM showing at the bottom basaloid cells. Pyknotic nuclei and perinuclear halo are seen in transitional cells (H&E, 9400). (B) PM – Eosinophilic mass of GC characterized by absence of the nuclei and abundant eosinophilic cytoplasm (H&E, 9400). (C) CP – Numerous GC forming masses adjacent to cuboidal basal cells. Viable cells still presenting nuclei are surrounding some GC (H&E, 9400). (D) CCOT – Cuboidal basal cells and transitional cells at various stages of transformation to GC (HE, 9400). J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
3
A
B
C
D
Figure 2 Immunohistochemical expression of cytokeratin AE1/AE3 in PM, CP, and CCOT. (A) PM – Absence of expression is seen in basaloid cells and weak positivity in transitional and ghost cells (IHC, 9400). (B) PM – Sheet of GC weakly positive. Negative central round areas correspond to the space previously occupied by the nuclei (IHC, 9400). (C) CP – Positivity in the basal layer, stronger expression in transitional cells and very weak in GC (IHC, 9400). (D) CCOT showing expression in basal and transitional cells, while GC vary from negative to weakly positive (IHC, 9400).
surrounded by a clear halo (Fig. 1A). Transitional cells gradually transformed into eosinophilic masses of confluent GC, characterized by absence of nuclei (Fig. 1B). In some cases, masses of GC formed the bulk of the tumor. Craniopharyngiomas presented solid, cystic, and mixed patterns. Most of the cases (64%) were of the mixed type, showing cords and lobules of stellate reticulum-like epithelium, bordered by a basal layer with ameloblastic features. Isolated groups of GC were seen thorough the epithelial layer (Fig. 1C). Most cases of CCOT were cystic, lined by palisaded cuboidal to columnar basal cells. Upper layers of the cystic lining were composed of cells resembling stellate reticulum, with variable number of GC which could either form solid masses or remain isolated (Fig. 1D).
By immunohistochemistry, GC of PM, CP, and CCOT were in most cases weakly positive for AE1/AE3 (Fig. 2). Immunohistochemical findings of all the antibodies studied are briefly described here and summarized in Table 2. Transitional cells showed strong positivity for AE1/AE3 in CP and CCOT (Fig. 2C,D). Considering the other markers, in PM, CD138 and FAS showed weak to moderate expression. In CP, most of the cases expressed Glut-1, while b-catenin was found varying in pattern from cytoplasmic (25%) to nuclear (75%). Most cases of CCOT were moderately positive for b-catenin, and strongly expressed Glut-1 (Fig. 3). Basaloid cells of PM showed weak focal positivity for AE1/AE3, but it was negative for CKs 8, 14, and 19. Most
Table 2 Immunohistochemical expression in the different cell types of the tumors studied PM
CP
CCOT
Antibody
GC
TC
BC
GC
TC
BC
GC
TC
BC
AE1/AE3 CK8 CK14 CK19 CD138 b-catenin D2-40 Glut-1 FAS CD10
71% Neg Neg Neg Neg Neg Neg Neg 5% Neg
58% Neg Neg Neg 33% Neg Neg Neg 67% Neg
83% 83% Neg Neg 100% 83% Neg Neg 83% 100%
86% Neg Neg Neg Neg Neg Neg Neg 14% Neg
100% 46% 77% 77% 75% 50% Neg 92% 85% 75%
100% 57% 100% 21% 14% 7% 71% Neg 13% Neg
100% Neg Neg Neg Neg Neg Neg Neg 13% Neg
100% 38% 100% 100% 75% 88% 88% 75% 100% 100%
100% 38% 88% 75% 25% 88% 63% Neg 63% Neg
PM, pilomatrixoma; CP, craniopharyngioma; CCOT, calcifying cystic odontogenic tumor, GC, ghost cells; TC, transitional cells; BC, basaloid/basal cells. J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
4
A
B
C
D
E
F
Figure 3 Immunohistochemical expression of b-catenin and Glut-1 in PM, CP, and CCOT. (A) PM – Strong membranous b-catenin expression restricted to basaloid cells. Note that transitional cells are negative (IHC, 9400). (B) PM – Glut-1 positivity in focal squamous cell aggregates adjacent to basaloid cells (IHC, 9400). (C) CP – Nuclear and cytoplasmic b-catenin expression in transitional cells and in whorl-like arrays adjacent to clusters of GC (IHC 9400). (D) CP – Transitional cells adjacent to GC mass expressing membranous Glut-1 (IHC, 9400). (E) CCOT – b-catenin nuclear, cytoplasmic and membranous expression in transitional cells (IHC, 9400). (F) CCOT – Strong membranous expression of Glut-1 in transitional cells (IHC, 9400).
of the cases expressed b-catenin (mainly membrane) (Fig. 3A). Only one case was positive for b-catenin in the nuclei and cytoplasm. CCOT basal cells showed strong expression of AE1/AE3 (Fig. 2D), while b-catenin was positive mainly in the cytoplasm and membrane. When analyzed by SEM, GC of PM formed a sheet without evident cell borders, with the area corresponding to the cytoplasm showing a dense network of fibrils surrounding a central empty round space, where the nuclei was previously located (Fig. 4B). Transitional cells could be identified by the presence of a small central round structure, corresponding to the pyknotic nuclei seen in H&E (Fig. 4A). In CP and CCOT, GC showed similar morphology, but most of them were isolated or formed small clusters (Fig. 4C–E). GC exhibited again the cytoplasm filled with a network of irregular fibrils, leaving an empty central round space (Fig. 4D–F). J Oral Pathol Med
Discussion Pilomatrixomas are easily identified histologically by its unique features, that is, a thick basaloid basophilic layer, masses of GC, and a well-defined transitional zone between them. Ghost cells of PM show distinct morphologic features when compared to CP and CCOT, which present histopathologically GC with similar characteristics. In PM, GC form sheets which turn in solid large masses, while in CP and CCOT, these cells tend to form small clusters or to be isolated, finally forming solid sheets of variable sizes (12, 13). A transitional zone between basal cells and GC is clearly defined in PM, forming a layer immediately adjacent to basaloid cells, while in CP and CCOT, this layer is harder to identify. However, in CCOT, the cells adjacent to clusters or to isolated GC frequently exhibit pyknotic nuclei, which gradually disappear, and these cells correspond to an
Ghost cells study in three tumors Rumayor et al.
5
A
B
C
D
E
F
Figure 4 Scanning electron microscopy aspects of PM, CP, and CCOT. (A) PM – Transitional cells showing central small structures correspondent to the pyknotic nuclei (SEM, 9500). (B) PM – Mass of GC showing a net of cytoplasmic fibrils and round well-defined empty spaces corresponding to the previously nuclei (SEM, 9500). (C) CP – Groups of GC separated by transitional cells (SEM 9400). (D) CP – Higher magnification of a round GC showing an empty central area, and a network of multiple irregular fibrils filling the cytoplasm. Cells on the upper portion of the image are fused forming a small confluent mass (SEM 92500). (E) CCOT – GC seen showing an empty space centrally permeated by transitional cells (SEM, 9430). (F) Higher magnification of E showing GC separated by artifacts from the adjacent transitional cells. At the right side, two individual GC are fused (SEM, 92500).
intermediate stage to finally form GC. These intermediate cells also have a distinct immunohistochemical pattern, including expression of nuclear b-catenin in CP and CCOT, indicating a possible activation of the Wnt pathway, as described in GC of odontomas (3). Nevertheless, PM showed a different b-catenin pattern, a distinct strong membrane expression in basaloid cells, while it was negative in transitional cells (14, 15). In addition to bcatenin, transitional cells of CP and CCOT also share expression of other markers including CKs, CD138, Glut-1, FAS, and CD10, while in PM, most of these markers were negative in transitional cells. Glut-1 expression, which was restricted to the viable cells immediately adjacent to GC in CP and CCOT, has not been studied previously in these neoplasms. As metabolism probably decreases during GC transformation, Glut-1 expression on transitional cells could be related to hypoxia.
Regarding to the basal layer in CP and CCOT, and basaloid cells in PM, differences in immunohistochemical expression were observed. Interestingly and unlike Broekaert et al. (16) description, basaloid cells in PM were weakly and focally positive for AE1/AE3 and negative for CKs 8, 14, and 19. Other antibodies described in basaloid cells of PM include cell adhesion protein CD138, which in our study was positive in all the cases, showing a gradual loss of expression as transforming into GC (16, 17). CD10, a marker described in dermal follicles, and other skin neoplasms were strongly positive in basaloid cells of all PM cases (18). Differently to PM and according to the literature, the basal layer of CP and CCOT expressed various types of CKs, while CD138 was expressed only in a few cases (14–25%) as reported recently by Gomes da Silva et al. (19) in which 35% of CCOT of their study were positive in the basal layer. J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
6
Expression of CKs in GC of PM has not been described previously, and our cases were weakly positive for AE1/ AE3, similar to CP and CCOT. Kato et al. (6) described positivity in GC of CP only for the AE3 clone, while in CCOT besides AE1/AE3, weak positivity for 34bE12 and CK13 have also been described (5, 6, 20). Ghost cells in all our cases of PM, CP, and CCOT were negative for CKs 8, 14, and 19. Other types of proteins have been reported as being positive in GC, such as human hair proteins in PM, CP, CCOT, and odontomas (3, 7). Although this could be explained by the hair matrix cell origin of PM, it is more difficult to understand such positivity in CP and odontogenic tumors, but it was suggested that could be related to the Wnt/b-catenin pathway, linked to hair keratin genes and also with tooth development (21, 22). Basal and transitional cells, as they are viable, express various proteins besides CKs, but they are negative in GC. The expression of these proteins varies in PM, CP, and CCOT as shown in this work, and they can eventually be useful to understand the mechanisms involved in GC formation. By transmission electron microscopy, GC of CCOT showed bundles of tonofilaments, usually interpreted as keratin, and absence of organelles (9, 13, 23). In PM, the ultrastructural findings were similar in GC, and when compared with basaloid cells, the former showed an increase in size and number of the irregularly disposed filaments, that fused together forming a band around the central nuclear empty space (8, 24). Ultrastructural analyses in CP are scarce, but similar findings were described; tightly packed fibrillar structures and absence of cytoplasmic organelles (25). We observed by SEM that GC in CCOT and CP presented similar patterns, including well-defined outlines, interlacing fibrils surrounding a central empty round space, as described previously in GC of decalcified sections of complex odontomas (26). The CKs are the only fibrils described in GC, but its precise nature and characteristics are unknown. According to some authors, it does not represent normal or true keratin, as the immunohistochemical expression is weak and negative for most specific subtypes of CKs, and the term ‘aberrant keratinization’ has been suggested (26). More recently, based on immunohistochemical studies of PM, the term ‘apoptosis-like process’ has been used, referring to keratinization of GC (17). It is interesting that GC commonly maintain their contour, different from true keratinizing cells, which usually collapse into a flattened shape due to the aggregation of keratin filaments, promoted by filagrin (27). In short, the cytoplasmatic fibrillar structure found in GC of PM, CP, and CCOT represent tonofilaments of abnormal keratin, not yet well characterized.
Conclusions In summary, GC are typically found in PM, CP, and CCOT, and they are formed mainly by modified keratins by not yet known mechanisms. In PM, GC form mainly amorphous masses, while CP and CCOT frequently show small clusters or isolated cells of well-defined contour. By immunohistochemistry, protein expression was not uniform between the three tumors; despite their very similar microscopic J Oral Pathol Med
appearance, probably different pathways may be involved in their formation.
References 1. Li TJ, Yu SF. Clinicopathologic spectrum of the so-called calcifying odontogenic cysts: a study of 21 intraosseous cases with reconsideration of the terminology and classification. Am J Surg Pathol 2003; 27: 372–84. 2. Tomich CE. Calcifying odontogenic cyst and dentinogenic ghost cell tumor. Oral Maxillofacial Surg Clin N Am 2004; 16: 391–7. 3. Tanaka A, Okamoto M, Yoshizawa D, et al. Presence of ghost cells and the Wnt signaling pathway in odontomas. J Oral Pathol Med 2007; 36: 400–4. 4. Takata T, Zhao M, Nikai H, Uchida T, Wang T. Ghost cells in calcifying odontogenic cysts express enamel-related proteins. Histochem J 2000; 32: 223–9. 5. Fregnani ER, Pires FR, Quezada RD, Shih IEM, Vargas PA, de Almeida OP. Calcifying odontogenic cyst: clinicopathological features and immunohistochemical profile of 10 cases. J Oral Pathol Med 2003; 32: 163–70. 6. Kato K, Nakatani Y, Kanno H, et al. Possible linkage between specific histological structures and aberrant reactivation of the Wnt pathway in adamantinomatous craniopharyngioma. J Pathol 2004; 203: 814–21. 7. Kusama K, Katayama Y, Oba K, et al. Expression of hard alpha-keratins in pilomatrixoma, craniopharyngioma, and calcifying odontogenic cyst. Am J Clin Pathol 2005; 123: 376–81. 8. Hashimoto K, Nelson RG, Lever WF. Calcifying epithelioma of Malherbe. Histochemical and electron microscopic studies. J Invest Dermatol 1966; 46: 391–408. 9. Regezi JA, Courtney RM, Kerr DA. Keratinization in odontogenic tumors. Oral Surg Oral Med Oral Pathol 1975; 39: 447–55. 10. Chen SY, Miller AS. Ultrastructure of the keratinizing and calcifying odontogenic cyst. Oral Surg Oral Med Oral Pathol 1975; 39: 769–80. 11. Satomura K, Nakanishi H, Fujisawa K, Hayashi E, Nagayama M. Initiation of ectopic epithelial calcification in a calcifying odontogenic cyst. J Oral Pathol Med 1999; 28: 330–5. 12. Kaddu S, Soyer HP, H€ odl S, Kerl H. Morphological stages of pilomatricoma. Am J Dermatopathol 1996; 18: 333–8. 13. Lucchese A, Scivetti M, Pilolli GP, Favia G. Analysis of ghost cells in calcifying cystic odontogenic tumors by confocal laser scanning microscopy. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2007; 104: 391–4. 14. Buslei R, Nolde M, Hofmann B, et al. Common mutations of beta-catenin in adamantinomatous craniopharyngiomas but not in other tumours originating from the sellar region. Acta Neuropathol 2005; 109: 589–97. 15. Kim SA, Ahn SG, Kim SG, et al. Investigation of the betacatenin gene in a case of dentinogenic ghost cell tumor. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2007; 103: 97–101. 16. Broekaert D, Goeman L, Ramaekers FC, et al. An investigation of cytokeratin expression in skin epithelial cysts and some uncommon types of cystic tumours using chainspecific antibodies. Arch Dermatol Res 1990; 282: 383– 91. 17. Nakamura T. Comparative immunohistochemical analyses on the modes of cell death/keratinization in epidermal cyst, trichilemmal cyst, and pilomatricoma. Am J Dermatopathol 2011; 33: 78–83.
Ghost cells study in three tumors Rumayor et al.
18. Yada K, Kashima K, Daa T, Kitano S, Fujiwara S, Yokoyama S. Expression of CD10 in basal cell carcinoma. Am J Dermatopathol 2004; 26: 463–71. 19. Gomes da Silva W, Ribeiro Bartholomeu Dos Santos TC, Cabral MG, Azevedo RS, Pires FR. Clinicopathologic analysis and syndecan-1 and Ki-67 expression in calcifying cystic odontogenic tumors, dentinogenic ghost cell tumor, and ghost cell odontogenic carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol. 2014; 117: 626–33. 20. El-Sissy NA, Rashad NA. CK13 in craniopharyngioma versus related odontogenic neoplasms and human enamel organ. East Mediterr Health J 1999; 5: 490–502. 21. Moreno-Bueno G, Gamallo C, Perez-Gallego L, Contreras F, Palacios J. beta-catenin expression in pilomatrixomas. Relationship with beta-catenin gene mutations and comparison with beta-catenin expression in normal hair follicles. Br J Dermatol 2001; 145: 576–81. 22. Liu F, Chu EY, Watt B, et al. Wnt/beta-catenin signaling directs multiple stages of tooth morphogenesis. Dev Biol 2008; 313: 210–24. 23. Vuletin JC, Solomon MP, Pertschuk LP. Peripheral odontogenic tumor with ghost-cell keratinization. A histologic,
24. 25. 26. 27.
fluorescent microscopic, and ultrastructural study. Oral Surg Oral Med Oral Pathol 1978; 45: 406–15. McGavran MH. Ultrastructure of pilomatrixoma (calcifying epithelioma). Cancer 1965; 18: 1445–56. Ghatak NR, Hirano A, Zimmerman HN. Ultrastructure of a craniopharyngioma. Cancer 1971; 27: 1465–75. Kerebel B, Kerebel LM. Ghost cells in complex odontoma: a light microscopic and SEM study. Oral Surg Oral Med Oral Pathol 1985; 59: 371–8. Candi E, Schmidt R, Melino G. The cornified envelope: a model of cell death in the skin. Nat Rev Mol Cell Biol 2005; 6: 328–40.
7
Acknowledgements This work was supported by the Brazilian Federal Agency for the Support and Evaluation of Graduate Education (CAPES).
Conflict of interest There are no conflicts of interest or disclosures regarding to this work.
J Oral Pathol Med
J Oral Pathol Med © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd wileyonlinelibrary.com/journal/jop
Ghost cells in pilomatrixoma, craniopharyngioma, and calcifying cystic odontogenic tumor: histological, immunohistochemical, and ultrastructural study Alicia Rumayor1, Rom an Carlos2, Hern an Molina Kirsch3, Bruno A. Benevenuto de Andrade4, Mario J. 4 1 Roma~ nach , Oslei Paes de Almeida 1
Oral Pathology Section, Department of Oral Diagnosis, Piracicaba Dental School, University of Campinas/UNICAMP, Piracicaba, Brazil; 2Pathology Division, Centro Clınico de Cabeza y Cuello/Hospital Herrera Llerandi, Guatemala City, Guatemala; 3Pathology Laboratory, Guatemala City, Guatemala; 4Oral Pathology Section, Department of Oral Diagnosis and Pathology, School of Dentistry, Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, Brazil
BACKGROUND: Pilomatrixoma, craniopharyngioma, and calcifying cystic odontogenic tumor are the main entities presenting ghost cells as an important histological feature, in spite their quite different clinical presentation; it seems that they share a common pathway in the formation of these cells. The aim of this study is to examine and compare the characteristics of ghost and other cells that form these lesions. METHODS: Forty-three cases including 21 pilomatrixomas, 14 craniopharyngiomas, and eight calcifying cystic odontogenic tumors were evaluated by immunohistochemistry for cytokeratins, CD138, b-catenin, D2-40, Glut-1, FAS, CD10 and also by scanning electron microscopy. RESULTS: The CKs, CD138, b-catenin, Glut-1, FAS, and CD10 were more often expressed by transitional cells of craniopharyngioma and calcifying cystic odontogenic tumor, compared with pilomatrixoma. Basaloid cells of pilomatrixoma showed strong positivity for CD138 and CD10. Differences on expression pattern were identified in transitional and basal cells, as ghost cells were negative for most antibodies used, except by low expression for cytokeratins. By scanning electron microscopy, the morphology of ghost cells were similar in their fibrillar cytoplasm, but their pattern varied from sheets in pilomatrixoma to small clusters in craniopharyngioma and calcifying cystic odontogenic tumor. CONCLUSIONS: Mechanisms involved in formation of ghost cells are unknown, but probably they follow different pathways as protein expression in the basal/
Correspondence: Alicia Rumayor, DDS, MSc, Oral Pathology, Piracicaba Dental School, University of Campinas (UNICAMP), Av. Limeira 901, P.O. Box 52, 13414-903, Piracicaba, S~ao Paulo, Brazil. Tel: +551921065315, Fax: +551921065218, E-mail: [email protected] Accepted for publication June 17, 2014
transitional cells was not uniform in the three tumors studied. J Oral Pathol Med (2014) Keywords: calcifying cystic odontogenic tumor; craniopharyngioma; ghost cells; immunohistochemistry; pilomatrixoma
Introduction Ghost cells (GC) are characterized by their abundant eosinophilic cytoplasm, a central empty round space which corresponds to the area previously occupied by the nuclei and a well-defined outline. They are typically observed in calcifying cystic odontogenic tumor (CCOT), craniopharyngioma (CP), pilomatrixoma (PM) and eventually can also be found in ameloblastoma, odontoma and ameloblastic fibro-odontoma (1–3). Although the mechanisms involved in GC formation are unknown; most authors consider it an unusual or aberrant form of cellular keratinization, called ‘ghost cell keratinization’. By immunohistochemistry, CKs in GC vary in type and intensity. Besides, enamel-related proteins have been described in GC of CCOT, and human hair proteins in PM, CP and CCOT (4–7). Ultrastructural studies of PM have shown bundles of fibrils within the cytoplasm of GC, interpreted as tonofilaments, probably related to keratins. A similar fibrillar pattern was also described in GC of CCOT. The loss of the nucleus in the cytoplasm of GC generates an empty round central space identified as an electron-lucent area by transmission electron microscopy (TEM) (8–11). The objective of this study is to describe and compare the histopathological, immunohistochemical, and ultrastructural features of GC and its precursors in PM, CP, and CCOT.
Ghost cells study in three tumors Rumayor et al.
2
Table 1 Antibodies used for the immunohistochemical analysis
Antibody
Clone
Source
Dilution
AE1/AE3
AE1/AE3
Dako
1:500
CK8 CK14 CK19 CD138 b-catenin
TS1 LL002 RCK 108 My 15 17 C 2
Novocastra Novocastra Dako Dako Novocastra
1:500 1:200 1:200 1:100 1:200
D2-40 Glut-1 FAS CD10
D2-40 Polyclonal Polyclonal 56C6
Dako Bio Systems Atlas Novocastra
1:100 1:100 1:200 1:100
Antigen specificity Cocktail/Epithelial cell Epithelial cell Epithelial cell Epithelial cell Intercellular junction Intercellular junction Wnt signaling pathway Basal epithelial cells Glucose transporter Fatty acid synthesis Cell surface metalloproteinase
Materials and methods Forty-three cases including 21 PM, 14 CP, and eight CCOT were used in this study. Histological 3-lm-thick sections of paraffin-embedded tissue were stained with H&E and processed for conventional immunohistochemistry and scanning electron microscopy (SEM). For immunohistochemistry, sections were dewaxed with xylene and then hydrated in ethanol. After antigen retrieval, endogenous peroxidase activity was blocked using five baths of 10% hydrogen peroxide, 5 min each. Antigen retrieval was performed in a pressure cooker with citrate buffer (pH
6.0), except for FAS and CD10 in which EDTA/Tris (pH 9.0) was used. After being washed in phosphate-buffered saline (pH 7.4), slides were incubated overnight with primary antibodies (Table 1). All slides were subsequently exposed to avidin–biotin complex and peroxidase reagents (LSAB kit; Dako Cytomation, Glostrup, Denmark) and diaminobenzidine tetrahydrochloride (Sigma, St. Louis, MO, USA), and subsequently counterstained with Carazzi hematoxylin. Adequate positive control sections were used for each antibody, and the negative control was obtained by omitting the primary specific antibody. A descriptive analysis of histopathological and immunohistochemical findings was performed for all the markers, considering GC and its precursors. For SEM, 7-lm-thick sections of paraffin-embedded tissue were dewaxed with xylene, dried with ethanol, then mounted on a metallic stub, and coated with a thin layer of gold. Images were taken at a voltage of 15 kV with magnifications from 400 to 2500 times. The current study was approved by the Ethical Committee of Piracicaba Dental School, State University of Campinas.
Results Histopathologically, 90% of PM presented a solid pattern, with only two cases showing cystic features. Ten and nine cases of solid PM were classified as early and late regressive stage, respectively. Basaloid cells showed intensely basophilic nuclei, scanty cytoplasm, and indistinct cellular borders, a distinct transitional region between basaloid and GC was evident, formed by cells with a pyknotic nuclei
A
B
C
D
Figure 1 Histopathological aspects of PM, CP, and CCOT. (A) PM showing at the bottom basaloid cells. Pyknotic nuclei and perinuclear halo are seen in transitional cells (H&E, 9400). (B) PM – Eosinophilic mass of GC characterized by absence of the nuclei and abundant eosinophilic cytoplasm (H&E, 9400). (C) CP – Numerous GC forming masses adjacent to cuboidal basal cells. Viable cells still presenting nuclei are surrounding some GC (H&E, 9400). (D) CCOT – Cuboidal basal cells and transitional cells at various stages of transformation to GC (HE, 9400). J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
3
A
B
C
D
Figure 2 Immunohistochemical expression of cytokeratin AE1/AE3 in PM, CP, and CCOT. (A) PM – Absence of expression is seen in basaloid cells and weak positivity in transitional and ghost cells (IHC, 9400). (B) PM – Sheet of GC weakly positive. Negative central round areas correspond to the space previously occupied by the nuclei (IHC, 9400). (C) CP – Positivity in the basal layer, stronger expression in transitional cells and very weak in GC (IHC, 9400). (D) CCOT showing expression in basal and transitional cells, while GC vary from negative to weakly positive (IHC, 9400).
surrounded by a clear halo (Fig. 1A). Transitional cells gradually transformed into eosinophilic masses of confluent GC, characterized by absence of nuclei (Fig. 1B). In some cases, masses of GC formed the bulk of the tumor. Craniopharyngiomas presented solid, cystic, and mixed patterns. Most of the cases (64%) were of the mixed type, showing cords and lobules of stellate reticulum-like epithelium, bordered by a basal layer with ameloblastic features. Isolated groups of GC were seen thorough the epithelial layer (Fig. 1C). Most cases of CCOT were cystic, lined by palisaded cuboidal to columnar basal cells. Upper layers of the cystic lining were composed of cells resembling stellate reticulum, with variable number of GC which could either form solid masses or remain isolated (Fig. 1D).
By immunohistochemistry, GC of PM, CP, and CCOT were in most cases weakly positive for AE1/AE3 (Fig. 2). Immunohistochemical findings of all the antibodies studied are briefly described here and summarized in Table 2. Transitional cells showed strong positivity for AE1/AE3 in CP and CCOT (Fig. 2C,D). Considering the other markers, in PM, CD138 and FAS showed weak to moderate expression. In CP, most of the cases expressed Glut-1, while b-catenin was found varying in pattern from cytoplasmic (25%) to nuclear (75%). Most cases of CCOT were moderately positive for b-catenin, and strongly expressed Glut-1 (Fig. 3). Basaloid cells of PM showed weak focal positivity for AE1/AE3, but it was negative for CKs 8, 14, and 19. Most
Table 2 Immunohistochemical expression in the different cell types of the tumors studied PM
CP
CCOT
Antibody
GC
TC
BC
GC
TC
BC
GC
TC
BC
AE1/AE3 CK8 CK14 CK19 CD138 b-catenin D2-40 Glut-1 FAS CD10
71% Neg Neg Neg Neg Neg Neg Neg 5% Neg
58% Neg Neg Neg 33% Neg Neg Neg 67% Neg
83% 83% Neg Neg 100% 83% Neg Neg 83% 100%
86% Neg Neg Neg Neg Neg Neg Neg 14% Neg
100% 46% 77% 77% 75% 50% Neg 92% 85% 75%
100% 57% 100% 21% 14% 7% 71% Neg 13% Neg
100% Neg Neg Neg Neg Neg Neg Neg 13% Neg
100% 38% 100% 100% 75% 88% 88% 75% 100% 100%
100% 38% 88% 75% 25% 88% 63% Neg 63% Neg
PM, pilomatrixoma; CP, craniopharyngioma; CCOT, calcifying cystic odontogenic tumor, GC, ghost cells; TC, transitional cells; BC, basaloid/basal cells. J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
4
A
B
C
D
E
F
Figure 3 Immunohistochemical expression of b-catenin and Glut-1 in PM, CP, and CCOT. (A) PM – Strong membranous b-catenin expression restricted to basaloid cells. Note that transitional cells are negative (IHC, 9400). (B) PM – Glut-1 positivity in focal squamous cell aggregates adjacent to basaloid cells (IHC, 9400). (C) CP – Nuclear and cytoplasmic b-catenin expression in transitional cells and in whorl-like arrays adjacent to clusters of GC (IHC 9400). (D) CP – Transitional cells adjacent to GC mass expressing membranous Glut-1 (IHC, 9400). (E) CCOT – b-catenin nuclear, cytoplasmic and membranous expression in transitional cells (IHC, 9400). (F) CCOT – Strong membranous expression of Glut-1 in transitional cells (IHC, 9400).
of the cases expressed b-catenin (mainly membrane) (Fig. 3A). Only one case was positive for b-catenin in the nuclei and cytoplasm. CCOT basal cells showed strong expression of AE1/AE3 (Fig. 2D), while b-catenin was positive mainly in the cytoplasm and membrane. When analyzed by SEM, GC of PM formed a sheet without evident cell borders, with the area corresponding to the cytoplasm showing a dense network of fibrils surrounding a central empty round space, where the nuclei was previously located (Fig. 4B). Transitional cells could be identified by the presence of a small central round structure, corresponding to the pyknotic nuclei seen in H&E (Fig. 4A). In CP and CCOT, GC showed similar morphology, but most of them were isolated or formed small clusters (Fig. 4C–E). GC exhibited again the cytoplasm filled with a network of irregular fibrils, leaving an empty central round space (Fig. 4D–F). J Oral Pathol Med
Discussion Pilomatrixomas are easily identified histologically by its unique features, that is, a thick basaloid basophilic layer, masses of GC, and a well-defined transitional zone between them. Ghost cells of PM show distinct morphologic features when compared to CP and CCOT, which present histopathologically GC with similar characteristics. In PM, GC form sheets which turn in solid large masses, while in CP and CCOT, these cells tend to form small clusters or to be isolated, finally forming solid sheets of variable sizes (12, 13). A transitional zone between basal cells and GC is clearly defined in PM, forming a layer immediately adjacent to basaloid cells, while in CP and CCOT, this layer is harder to identify. However, in CCOT, the cells adjacent to clusters or to isolated GC frequently exhibit pyknotic nuclei, which gradually disappear, and these cells correspond to an
Ghost cells study in three tumors Rumayor et al.
5
A
B
C
D
E
F
Figure 4 Scanning electron microscopy aspects of PM, CP, and CCOT. (A) PM – Transitional cells showing central small structures correspondent to the pyknotic nuclei (SEM, 9500). (B) PM – Mass of GC showing a net of cytoplasmic fibrils and round well-defined empty spaces corresponding to the previously nuclei (SEM, 9500). (C) CP – Groups of GC separated by transitional cells (SEM 9400). (D) CP – Higher magnification of a round GC showing an empty central area, and a network of multiple irregular fibrils filling the cytoplasm. Cells on the upper portion of the image are fused forming a small confluent mass (SEM 92500). (E) CCOT – GC seen showing an empty space centrally permeated by transitional cells (SEM, 9430). (F) Higher magnification of E showing GC separated by artifacts from the adjacent transitional cells. At the right side, two individual GC are fused (SEM, 92500).
intermediate stage to finally form GC. These intermediate cells also have a distinct immunohistochemical pattern, including expression of nuclear b-catenin in CP and CCOT, indicating a possible activation of the Wnt pathway, as described in GC of odontomas (3). Nevertheless, PM showed a different b-catenin pattern, a distinct strong membrane expression in basaloid cells, while it was negative in transitional cells (14, 15). In addition to bcatenin, transitional cells of CP and CCOT also share expression of other markers including CKs, CD138, Glut-1, FAS, and CD10, while in PM, most of these markers were negative in transitional cells. Glut-1 expression, which was restricted to the viable cells immediately adjacent to GC in CP and CCOT, has not been studied previously in these neoplasms. As metabolism probably decreases during GC transformation, Glut-1 expression on transitional cells could be related to hypoxia.
Regarding to the basal layer in CP and CCOT, and basaloid cells in PM, differences in immunohistochemical expression were observed. Interestingly and unlike Broekaert et al. (16) description, basaloid cells in PM were weakly and focally positive for AE1/AE3 and negative for CKs 8, 14, and 19. Other antibodies described in basaloid cells of PM include cell adhesion protein CD138, which in our study was positive in all the cases, showing a gradual loss of expression as transforming into GC (16, 17). CD10, a marker described in dermal follicles, and other skin neoplasms were strongly positive in basaloid cells of all PM cases (18). Differently to PM and according to the literature, the basal layer of CP and CCOT expressed various types of CKs, while CD138 was expressed only in a few cases (14–25%) as reported recently by Gomes da Silva et al. (19) in which 35% of CCOT of their study were positive in the basal layer. J Oral Pathol Med
Ghost cells study in three tumors Rumayor et al.
6
Expression of CKs in GC of PM has not been described previously, and our cases were weakly positive for AE1/ AE3, similar to CP and CCOT. Kato et al. (6) described positivity in GC of CP only for the AE3 clone, while in CCOT besides AE1/AE3, weak positivity for 34bE12 and CK13 have also been described (5, 6, 20). Ghost cells in all our cases of PM, CP, and CCOT were negative for CKs 8, 14, and 19. Other types of proteins have been reported as being positive in GC, such as human hair proteins in PM, CP, CCOT, and odontomas (3, 7). Although this could be explained by the hair matrix cell origin of PM, it is more difficult to understand such positivity in CP and odontogenic tumors, but it was suggested that could be related to the Wnt/b-catenin pathway, linked to hair keratin genes and also with tooth development (21, 22). Basal and transitional cells, as they are viable, express various proteins besides CKs, but they are negative in GC. The expression of these proteins varies in PM, CP, and CCOT as shown in this work, and they can eventually be useful to understand the mechanisms involved in GC formation. By transmission electron microscopy, GC of CCOT showed bundles of tonofilaments, usually interpreted as keratin, and absence of organelles (9, 13, 23). In PM, the ultrastructural findings were similar in GC, and when compared with basaloid cells, the former showed an increase in size and number of the irregularly disposed filaments, that fused together forming a band around the central nuclear empty space (8, 24). Ultrastructural analyses in CP are scarce, but similar findings were described; tightly packed fibrillar structures and absence of cytoplasmic organelles (25). We observed by SEM that GC in CCOT and CP presented similar patterns, including well-defined outlines, interlacing fibrils surrounding a central empty round space, as described previously in GC of decalcified sections of complex odontomas (26). The CKs are the only fibrils described in GC, but its precise nature and characteristics are unknown. According to some authors, it does not represent normal or true keratin, as the immunohistochemical expression is weak and negative for most specific subtypes of CKs, and the term ‘aberrant keratinization’ has been suggested (26). More recently, based on immunohistochemical studies of PM, the term ‘apoptosis-like process’ has been used, referring to keratinization of GC (17). It is interesting that GC commonly maintain their contour, different from true keratinizing cells, which usually collapse into a flattened shape due to the aggregation of keratin filaments, promoted by filagrin (27). In short, the cytoplasmatic fibrillar structure found in GC of PM, CP, and CCOT represent tonofilaments of abnormal keratin, not yet well characterized.
Conclusions In summary, GC are typically found in PM, CP, and CCOT, and they are formed mainly by modified keratins by not yet known mechanisms. In PM, GC form mainly amorphous masses, while CP and CCOT frequently show small clusters or isolated cells of well-defined contour. By immunohistochemistry, protein expression was not uniform between the three tumors; despite their very similar microscopic J Oral Pathol Med
appearance, probably different pathways may be involved in their formation.
References 1. Li TJ, Yu SF. Clinicopathologic spectrum of the so-called calcifying odontogenic cysts: a study of 21 intraosseous cases with reconsideration of the terminology and classification. Am J Surg Pathol 2003; 27: 372–84. 2. Tomich CE. Calcifying odontogenic cyst and dentinogenic ghost cell tumor. Oral Maxillofacial Surg Clin N Am 2004; 16: 391–7. 3. Tanaka A, Okamoto M, Yoshizawa D, et al. Presence of ghost cells and the Wnt signaling pathway in odontomas. J Oral Pathol Med 2007; 36: 400–4. 4. Takata T, Zhao M, Nikai H, Uchida T, Wang T. Ghost cells in calcifying odontogenic cysts express enamel-related proteins. Histochem J 2000; 32: 223–9. 5. Fregnani ER, Pires FR, Quezada RD, Shih IEM, Vargas PA, de Almeida OP. Calcifying odontogenic cyst: clinicopathological features and immunohistochemical profile of 10 cases. J Oral Pathol Med 2003; 32: 163–70. 6. Kato K, Nakatani Y, Kanno H, et al. Possible linkage between specific histological structures and aberrant reactivation of the Wnt pathway in adamantinomatous craniopharyngioma. J Pathol 2004; 203: 814–21. 7. Kusama K, Katayama Y, Oba K, et al. Expression of hard alpha-keratins in pilomatrixoma, craniopharyngioma, and calcifying odontogenic cyst. Am J Clin Pathol 2005; 123: 376–81. 8. Hashimoto K, Nelson RG, Lever WF. Calcifying epithelioma of Malherbe. Histochemical and electron microscopic studies. J Invest Dermatol 1966; 46: 391–408. 9. Regezi JA, Courtney RM, Kerr DA. Keratinization in odontogenic tumors. Oral Surg Oral Med Oral Pathol 1975; 39: 447–55. 10. Chen SY, Miller AS. Ultrastructure of the keratinizing and calcifying odontogenic cyst. Oral Surg Oral Med Oral Pathol 1975; 39: 769–80. 11. Satomura K, Nakanishi H, Fujisawa K, Hayashi E, Nagayama M. Initiation of ectopic epithelial calcification in a calcifying odontogenic cyst. J Oral Pathol Med 1999; 28: 330–5. 12. Kaddu S, Soyer HP, H€ odl S, Kerl H. Morphological stages of pilomatricoma. Am J Dermatopathol 1996; 18: 333–8. 13. Lucchese A, Scivetti M, Pilolli GP, Favia G. Analysis of ghost cells in calcifying cystic odontogenic tumors by confocal laser scanning microscopy. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2007; 104: 391–4. 14. Buslei R, Nolde M, Hofmann B, et al. Common mutations of beta-catenin in adamantinomatous craniopharyngiomas but not in other tumours originating from the sellar region. Acta Neuropathol 2005; 109: 589–97. 15. Kim SA, Ahn SG, Kim SG, et al. Investigation of the betacatenin gene in a case of dentinogenic ghost cell tumor. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2007; 103: 97–101. 16. Broekaert D, Goeman L, Ramaekers FC, et al. An investigation of cytokeratin expression in skin epithelial cysts and some uncommon types of cystic tumours using chainspecific antibodies. Arch Dermatol Res 1990; 282: 383– 91. 17. Nakamura T. Comparative immunohistochemical analyses on the modes of cell death/keratinization in epidermal cyst, trichilemmal cyst, and pilomatricoma. Am J Dermatopathol 2011; 33: 78–83.
Ghost cells study in three tumors Rumayor et al.
18. Yada K, Kashima K, Daa T, Kitano S, Fujiwara S, Yokoyama S. Expression of CD10 in basal cell carcinoma. Am J Dermatopathol 2004; 26: 463–71. 19. Gomes da Silva W, Ribeiro Bartholomeu Dos Santos TC, Cabral MG, Azevedo RS, Pires FR. Clinicopathologic analysis and syndecan-1 and Ki-67 expression in calcifying cystic odontogenic tumors, dentinogenic ghost cell tumor, and ghost cell odontogenic carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol. 2014; 117: 626–33. 20. El-Sissy NA, Rashad NA. CK13 in craniopharyngioma versus related odontogenic neoplasms and human enamel organ. East Mediterr Health J 1999; 5: 490–502. 21. Moreno-Bueno G, Gamallo C, Perez-Gallego L, Contreras F, Palacios J. beta-catenin expression in pilomatrixomas. Relationship with beta-catenin gene mutations and comparison with beta-catenin expression in normal hair follicles. Br J Dermatol 2001; 145: 576–81. 22. Liu F, Chu EY, Watt B, et al. Wnt/beta-catenin signaling directs multiple stages of tooth morphogenesis. Dev Biol 2008; 313: 210–24. 23. Vuletin JC, Solomon MP, Pertschuk LP. Peripheral odontogenic tumor with ghost-cell keratinization. A histologic,
24. 25. 26. 27.
fluorescent microscopic, and ultrastructural study. Oral Surg Oral Med Oral Pathol 1978; 45: 406–15. McGavran MH. Ultrastructure of pilomatrixoma (calcifying epithelioma). Cancer 1965; 18: 1445–56. Ghatak NR, Hirano A, Zimmerman HN. Ultrastructure of a craniopharyngioma. Cancer 1971; 27: 1465–75. Kerebel B, Kerebel LM. Ghost cells in complex odontoma: a light microscopic and SEM study. Oral Surg Oral Med Oral Pathol 1985; 59: 371–8. Candi E, Schmidt R, Melino G. The cornified envelope: a model of cell death in the skin. Nat Rev Mol Cell Biol 2005; 6: 328–40.
7
Acknowledgements This work was supported by the Brazilian Federal Agency for the Support and Evaluation of Graduate Education (CAPES).
Conflict of interest There are no conflicts of interest or disclosures regarding to this work.
J Oral Pathol Med
