The Amyloid Deposit in Calcifying Epithelial Odontogenic Tumor Is Immunoreactive for Cytokeratins Open Access

Since Pindborg first described calcifying epithelial odontogenic tumor (CEOT) as an entity in 1955, attempts have been made to understand the nature of the amyloid-like material present in this tumor and to find its origin. Presumably, the amyloid is a degeneration product of the tumor's epithelial cells. However, controversy still exists as to whether the amyloid is a degeneration product or an actively secreted material, and whether it is extracellular or intracellular in origin.1 

In this article, we present a new case of CEOT with an emphasis on the amyloid study, and we review the literature with respect to the histopathologic aspects of CEOT.

A 62-year-old woman was admitted to an outpatient clinic with a 2-month history of right nasal obstruction. Following identification of an obstructive nasal mass, a biopsy was taken and a diagnosis of anaplastic carcinoma was made. The patient was referred to the Department of Ear, Nose, and Throat at the Sheba Medical Center (Tel-Hashomer, Israel) for further evaluation. Physical examination was positive only for a nasal mass, and no neural deficit, trismus, or ophthalmic pathology was noted. Radiological study revealed that the right maxillary sinus was totally obliterated by a multilocular mass containing numerous radiopacities within the radiolucent areas.

The patient was offered a surgical treatment followed by radiation if indicated. The patient declined any form of treatment and was lost to follow-up. However, 8 months later the patient returned to the Head and Neck Clinic at the Sheba Medical Center after the failure of a radiation treatment of 7000 cGy, which she received at another institution. Physical examination revealed postradiation changes of the skin overlying the right facial area. The nasal mass was unchanged, and the rest of the physical examination was negative for tumor-related pathology.

The patient consented for surgery and underwent right maxillectomy with preservation of the orbital contents. The specimen was resected en bloc, and a temporary obturator prosthesis was inserted into the maxillary cavity. The patient's postoperative course was uneventful, and the patient was discharged on the sixth postoperative day.

Representative sections for light microscopy were fixed in 4% buffered formaldehyde and stained with hematoxylin-eosin, periodic acid–Schiff (PAS), mucicarmine, and Congo red. Immunohistochemical stains for laminin; collagen IV (Sigma Chemical Co, St Louis, Mo); amyloid-associated protein; prealbumin; a cocktail of cytokeratins 1, 5, 6, 8, 13, and 16 (Dakopatts A/S, Glostrup, Denmark); and cytokeratins AE1 and AE3 (Zymed Laboratories, South San Francisco, Calif) were performed. All cytokeratin stains were performed on a Ventana ES analyzer (Ventana Medical Systems Inc, Tucson, Ariz), using positive controls on each slide. The ultrastructural study was performed on formalin-fixed tissue, which was later fixed in glutaraldehyde. The amyloid was further assessed using the Shtrasburg method2,3 on formalin-fixed tissue.

The right maxilla, measuring 4.5 × 4.5 × 5 cm, was received for pathologic evaluation. Almost the entire maxilla was replaced by a firm mass that extended into the maxillary sinus and obliterated it completely.

Light microscopy revealed that the tumor was composed of polyhedral epithelial cells arranged in islands of glandlike structures and in some areas created sheets of closely packed cells. The cells had well-outlined borders and eosinophilic cytoplasm, and intercellular bridges could be recognized between them. Nuclei showed moderate to severe atypia with 1 or more prominent nucleoli, but no mitotic activity was evident. Extensive deposits of amorphous eosinophilic material were seen between the tumor cells and filled the glandlike structures. Numerous calcifications were found within the amorphous eosinophilic deposits, often showing multiple concentric hematoxyphilic rings, known as Liesegang ring phenomenon (Figure 1).

Stains for PAS, mucicarmine, laminin, and collagen IV were all negative. Tumor cells were cytoplasmically positive for all cytokeratin stains performed (cocktail, AE1, and AE3). The positivity for cytokeratins combined with the morphologic features of the cells, especially the presence of intercellular bridges, indicated that the tumor cells were of squamous origin. The eosinophilic material was positive for Congo red and was dichroic under polarized light.

The morphologic, histochemical, and immunohistochemical features of this tumor were consistent with CEOT, also known as Pindborg tumor.

The amyloid deposits were assessed further using different techniques. Immunohistochemical stains for amyloid-associated protein and prealbumin could not be interpreted as either positive or negative, since the entire tissue was weakly and diffusely stained. Amyloid analysis using the Shtrasburg method failed to show the presence of amyloid A. Ultrastructural study of the eosinophilic material showed 2 types of structures. The first and more predominant structure was a mesh of filaments measuring 10 to 15 nm in diameter, which is typical of amyloid. The fibrils were seen close to the epithelial cell membrane and microvilli, but not within the cells (Figure 2). The second structure, observed particularly around blood vessels, appeared as aggregates of basement membrane–like material and resembled lamina densa in structure.

Immunohistochemical stains for cytokeratins revealed the nature of the amyloid. Not only were the tumor cells positive for cytokeratins, but the amyloid itself was positive for all cytokeratin stains performed (cocktail, AE1, and AE3). The amyloid was not diffusely stained for cytokeratins, but rather stained focally, in a glandular-like pattern, reminiscent of the epithelial glandlike structures of the tumor. However, it was very clear that no epithelial cells existed in the stained area, which consisted only of amorphous eosinophilic material, positive for Congo red (Figure 3). To our knowledge, this positivity for cytokeratins has not been so vividly demonstrated previously.

Pindborg tumor is a rare, benign, but locally aggressive odontogenic tumor, which accounts for less than 1% of all odontogenic tumors. Most investigators believe that the tumor cells originate from the stratum intermedium of the normal dental germ. This idea is based on the morphologic similarity of the tumor cells to the normal cells of the stratum intermedium and on the finding of high activity of alkaline phosphatase and adenosine triphosphate at both sites.4 Calcifying epithelial odontogenic tumor occurs most commonly in the fourth and fifth decades; it has an age range of 8 to 92 years and a mean age at initial diagnosis of 40 years. The distribution among men and women is equal.1 There is a predilection for occurrence of the tumor in the mandible over the maxilla by a ratio of 2:1, and the prevalence in the molar region is 3 times that in the bicuspid region.1 Few extraosseous cases have been described, all of which have involved the gingiva.5,6 Fifty-two percent of the reported cases have been associated with an unerupted or embedded tooth or teeth.1 (In the case presented here, according to a repeated anamnesis, there was no tooth associated.) Clinically, most patients are aware only of a painless mass that is slowly growing. Some complain of nasal stuffiness, epistaxis, and headaches. Radiographic features are variable. The tumor most commonly appears as either a diffuse or a well-circumscribed unilocular radiolucent area. In some cases, the lesion becomes multilocular with a honeycomb pattern. In others, multiple radiopacities are seen within the radio-lucent area1 (similar to those seen in the case presented).

Histologically, the case we describe here represents a typical example of Pindborg tumor; it was composed of polyhedral epithelial cells that show distinct intercellular bridges, have considerable pleomorphism, and are associated with eosinophilic deposits with numerous calcifications.

A well-recognized form of this neoplasm is the clear cell variant, in which the tumor cells exhibit a clear vacuolated cytoplasm rather than an eosinophilic cytoplasm. Hybrid lesions with histologic features of both CEOT and adenodontogenic tumor have been reported.7–9 

The eosinophilic deposits mature with time, and as they mineralize they change from PAS-negative (as was in the presented case) to PAS-positive.1 At least in some cases this eosinophilic material was seen intracellularly. However, the question of degeneration product versus actively secreted substance, as well as the question of extracellular versus intracellular origin, are still matters of debate.1,10 

An ultrastructural study from 1984, carried out by El-Labban and colleges11 demonstrated that besides the polyhedral epithelial cells, there is another cell type present in CEOT, a cell with the ultrastructural characteristics of myoepithelial cells. These cells show a lamina densa that is continuous with the basal membrane of the tumor epithelial cells and also a large number of hemidesmosomes. This group also published the most recent ultrastructural study of the eosinophilic material.12 They revealed that the deposits consist of 2 types of structures, which are related. The first type appeared as sheets of fine filaments measuring 10 to 12 nm in diameter. The second type was in the form of aggregates of lamina densa fragments, probably secreted by the tumor epithelium. These fragments appeared to undergo some loss of electron density and became degraded into fine filaments, having similar thickness and electron density to those forming the filamentous masses. The authors concluded that the fine filamentous material is a form of amyloid that results from degradation of lamina densa material.12 However, earlier studies showed that the eosinophilic material, the amyloid, is composed of fibrils whose diameters vary, according to different publications, from 5 nm13–15 to 15 nm.13 This filamentous structure of the amyloid is comparable to that of other proteins, for example, keratin.1 Our findings support both the early and the more recent studies that identified amyloid within the eosinophilic material. We also repeat the observation of El-Labban's group that lamina densa material is present in addition to the amyloid. Yet, no evidence of the connection between the amyloid and the lamina densa material was found.

An immunohistochemical study of 4 cases of CEOT, conducted by Sauk et al,16 demonstrated the presence of basement membrane components (type IV collagen and laminin) in the amyloid deposits, unlike the results of the current study. In addition, Sauk and colleagues found a weak reaction for a 58-kd keratin protein in these extracellular deposits. Other molecular weight keratins were not examined.16 In a study by Mori et al,17 keratin was not detected in the amyloid using various antibodies against different molecular weight keratins.

In this work, cytokeratins were clearly identified, both in the tumor cells and in the amyloid, using several cytokeratin cocktails (1, 5, 6, 8, 13, and 16; AE1; and AE3). Two of these cocktail stains contain cytokeratin 5 (the first stain and AE3), which is specific for a 58-kd keratin protein. The facts that the stain was strong, unlike the findings of Sauk et al, and that the AE1 stain, which lacks cytokeratin 5, was also positive indicate that the amyloid contains more than 1 keratin protein.

There is no doubt that the epithelial cells of CEOT are of squamous epithelial origin, as is manifested by their morphologic, immunohistochemical, and ultrastructural characteristics.1 Amyloid deposit is well documented in association with squamous cell carcinoma at different locations, for example, the skin,18 vagina,19 uterine cervix,20 and nasopharynx.21 Some reports have indicated that squamous cell carcinoma–associated amyloid is immunoreactive for cytokeratins, and amyloid is regarded as the degradation product of keratin.18–20 It is also known that in primary cutaneous amyloidosis, the amyloid (known as amyloid K, meaning keratin or keratinoid amyloid) is derived from filamentous degeneration of keratin filaments.22,23 Therefore, it is reasonable to regard all these findings, including the CEOT amyloid, as similar expressions of the same process, that is, filamentous degeneration of keratin and conversion into amyloid. With this view, and since Pindborg tumor is an odontogenic tumor, one draws back to previous suggestions that the CEOT amyloid is somehow connected to enamel formation, which is related to keratin.1,10,24 

Another idea that emerges from analysis of the cytokeratin staining pattern of the amyloid is the concept of aging. This thought is based on several observations, among which is the resemblance between the cytokeratin staining pattern of the amyloid and the epithelial glandlike structures of the tumor. It seems that the tumor passes through various stages of development, commencing with epithelial degeneration and conversion of keratin filaments into amyloid. At first, this amyloid is PAS-negative and stains positively for cytokeratins. However, as it ages and the amyloid deposits coalesce into globules, the amyloid loses its immunoreactivity for cytokeratins and becomes PAS-positive. At this stage, the amyloid also mineralizes and gives rise to calcifications that exhibit the Liesegang ring phenomenon. The idea of aging has been proposed in the past,1,25 but it gains a new dimension now that it relies on more facts.

Last but not least, one cannot ignore the accumulating data indicating that CEOT eosinophilic deposits contain basement membrane components.12,16 However, the relative significance and contribution of keratin and basement membrane protein to the development of the CEOT-associated amyloid are still not clear. It is for future studies to reveal this and other secrets concerning this rare and unique tumor.

The authors thank S. Shtrasburg, BSc, for his help in investigating the amyloid by performing the Shtrasburg method.