Expression of N-Cadherin in Salivary Gland Tumors

Cadherins are a family of single-pass transmembrane glycoproteins which act as mediators in calcium-dependent cell-cell adhesion and play an essential role in development, cell polarity and tissues morphogenesis [1]. They are classified according to tissue patterns of expression: type E, epithelial-cell derived [2]; type P, placenta-derived [3], and type N, nerve-derived [4] are considered as ‘classical cadherins’. Cadherins have a large extracellular domain for intercellular homotypic connections and a cytoplasmic domain which interacts with cytoplasmic proteins called catenins (α-, β- and γ-catenin) that link to the actin-based cytoskeleton [5,6]. Catenins are critical for signal transduction and various cellular functions [5,7]. Normal epithelial cells typically express E-cadherin, whereas mesenchymal cells express N-cadherin [6].

N-cadherin was originally found in neural tissues [4], and later in smooth muscle cells [8], endothelial cells [9,10], fibroblasts [11], myofibroblasts [12], and mast cells [13]. In cancer cells, expression of E-cadherin at cell-cell connections acts as an invasion suppressor complex [14], whereas N-cadherin in malignant cells of epithelial origin promotes cellular motility, migration, invasion and metastasis [15,16]. N-cadherin has a dual role in the invasive process [17]. It promotes adhesive interactions among tumor cells with the microenvironment, and synergizes with the fibroblast growth factor receptor (FGFr) [18,19], thereby inducing the invasiveness of tumor cells.

Studies concerning the expression of N-cadherin in normal salivary glands and their benign-malignant neoplasms are only rarely referred to in the literature focusing only on specific subtype of salivary gland malignancies [20].

The purpose of this study was to detect immunohistochemically the expression of N-cadherin in different types of benign and malignant salivary gland tumors in an attempt to note any possible correlation with their development, stage and invasive properties.

Forty-nine tissue specimens of benign and malignant salivary gland tumors, formalin-fixed, and paraffin-embedded were retrieved from the archives of Department of Histopathology of ‘G. Papanikolaou’ General Hospital and ‘Theagenion’ Anticancer Hospital, Thessaloniki, Greece. Benign salivary gland tumors included pleomorphic adenomas (n = 5), Warthin’s tumors (n = 10), and myoepitheliomas (n = 4). Malignant salivary tumors included adenoid cystic carcinomas (ACC) (n = 14), mucoepidermoid carcinomas (MEC) (n = 4), polymorphous low-grade adenocarcinomas (PLGA) (n = 6), and adenocarcinomas not otherwise specified (NOS) (n = 5). The diagnosis of these cases was based on the established criteria of World Health Organisation (WHO) [21], and the stage of malignant tumors was obtained from patients’ medical records. The tumors were subdivided into two groups: local tumors (stages I–II) and invasive/metastatic tumors (stages III–IV). New 3-µm sections were cut and stained with hematoxylin and eosin and were reviewed for adequacy of tissue to study and agreement with the original diagnosis. Also, the sections from each specimen of malignant tumors were examined for the presence of perineural invasion. Twelve specimens of normal salivary tissue adjacent to benign neoplasms were used as control.

For the immunohistochemical detection of N-cadherin, antigen retrieval was performed including placement of the sections in a pressure cooker of 56°C for 24 h, following by microwave heat induced epitope retrieval for 20 min with Bond Epitope Retrieval Solution 1 (AR 9961, Vision BioSystems, Newcastle Upon Tyne, UK). Endogenous peroxidase activity was quenched with 3% H₂O₂ in methanol for 10 min and then the sections were rinsed with Tris buffer (BDH Lab Suppliers, Poole, UK). The Bond Polymer Refine Detection Kit (DS9800, Vision BioSystems) was applied using the mouse primary monoclonal antibody N-cadherin (M-3612, Dako, Glostrup, Denmark) at a dilution 1:100 for 30 min. The procedure was performed using the autostainer Bond-max processing module (code 21005, Vision BioSystems) according to the manufacturer’s instructions. Finally, the sections were washed in water and counterstained with hematoxylin. Sections incubated in Tris buffer and serum instead of the primary antibody were used as negative controls. The neural tissue of normal salivary glands was used as the internal positive control.

The localization, staining intensity and immunoreactivity were examined according to our previous study [22] as follows: the localization or staining pattern was classified as membranous and cytoplasmic when the immunostaining was localized in membrane and cytoplasm of cells, respectively. The staining intensity was assessed using the following brief evaluation: weak, moderate and strong. The sum of the staining intensity was used for the total immunoreactivity. The evaluation of immunoreactivity was performed as follows: 0 = negative or positive staining in ≤2% of cells; ± = positive staining in 3–10% of cells; + = positive staining in 11–50% of cells; ++ = positive staining in 51–70% of cells; +++ = positive staining in 70–90% of cells; ++++ = positive staining in 91–100% of cells. Five hundred cells from five fields on each section were enumerated as the percentage of cells expressing N-cadherin.

Perineural invasion and stage of malignant tumors in relation to N-cadherin expression were also evaluated. Sections were examined by two of the authors (A.E. and L.S.) independent of each other. Sections were re-examined when there was difference and agreement was reached by consensus.

For the statistical analysis the χ² test was used. Statistical significance was considered when p was <0.05.

The whole study was performed according to the Declaration of Helsinki II and furthermore, the permission of the relevant hospitals was given for the review of the medical records.

N-cadherin expression was not found in the epithelial cells of normal salivary glands. Immunoreactivity was seen in neural tissue, endothelial and smooth muscle cells of vessels and mast cells in the stroma. Also, in 5 cases focal immunostaining was observed along membranes of basal cells of normal oral squamous epithelium.

In the 20 benign salivary gland tumors, N-cadherin expression was not found in pleomorphic adenomas and myoepitheliomas. Unexpected immunoreactivity (++) was seen in membranes of epithelial cells in one case of Warthin’s tumor (fig. 1a). In this case, careful re-examination of hematoxylin and eosin sections did not showed epithelial atypia, mitotic figures, necrosis or infiltration around the tumor. Furthermore, according to the medical records, the 5-year follow-up of the patients did not show any recurrence or metastasis. Also expression of N-cadherin localized in membranes of centrocytes in germinal centers of lymphoid tissue was observed in 4 cases of Warthin’s tumor (fig. 1b).

The 29 malignant salivary gland tumors showed N-cadherin expression in 13 (45%) cases. The immunoreactivity and localization of expression in these positive cases are presented in table 1. The immunoreactivity in the different types of malignant tumors was almost similar. The staining intensity in all positive cases varied from moderate to strong. Regarding the relation between N-cadherin expression and tumor stage our results showed positive expression in 11/24 of cases in stages I–II, and 2/5 of stages III–IV. N-cadherin expression was not significantly correlated with tumors’ stage (χ² = 0.018, p < 0.2). Perineural invasion in relation to N-cadherin expression is presented in table 2. N-cadherin expression was found in 1 case of the total 15 cases where perineural invasion was absent, and in 12 cases of the total 14 cases with perineural invasion. Perineural invasion was significantly correlated with N-cadherin expression (χ² = 11.7, p < 0.0001).

In details, N-cadherin expression was observed in all histologic patterns (tubular, cribriform, solid) of ACCs (fig. 2a) and characteristically in tumor cells invading neural tissues and vessels (fig. 2b). All types of epithelial cells in MECs were positively stained for N-cadherin (fig. 2c). The solid, cribriform, glandular, ductular, tubular and single-file histologic patterns were found to be positive for N-cadherin in PLGAs, but the presence and proportions of these histologic patterns differed within the same case and among different cases (fig. 2d). Neoplastic cells in duct-like or glandular and solid patterns were also positive in all examined cases of NOS. Similar positivity was also observed in trabecular histologic pattern additionally presented in one case. N-cadherin expression was observed in all these histologic patterns of this tumor (fig. 2e, f).

In the current study, N-cadherin was not found in the normal salivary gland epithelium. However, others [9,10,13,20] had reported N-cadherin expression in membranes of basal cells in normal oral squamous epithelium, neural tissue, endothelial and smooth muscle cells of vessels, mast cells in the stroma and in the ductal cells of normal salivary glands. This difference could be explained by differences in immunohistochemical procedures [22] including the clonality of the primary antibody, the concentration of the primary and secondary antibodies, the length of incubation with such antibodies and the sensitivity of the enzymatic reaction [22].

In our study, N-cadherin expression was not observed in pleomorphic adenomas and myoepitheliomas but unexpectedly, N-cadherin expression was observed in membranes of epithelial cells in one case of Warthin’s tumor. Epithelial cells normally express E-cadherin, as reported in our previous study [23]. The N-cadherin expression in this case of Warthin’s tumor could be suggestive of a potential malignant transformation but careful re-examination of hematoxylin and eosin sections did not reveal epithelial atypia, mitotic figures, necrosis and infiltration of tissues adjacent to tumor. Furthermore, the five years that the patient was followed up did not show any recurrence or metastasis. Pathogenesis of Warthin’s tumor has long been controversial with regards to the origin of both epithelium and lymphoid stroma. Although several theories have been proposed, the theory involving development from heterotopic glandular tissue in a lymph node continues to be the most substantiated pathogenesis [24]. It is possible that expression of N-cadherin in epithelial cells of Warthin’s tumor to be transient during the development of heterotopic glandular tissue into lymphoid stroma with a process similar to that seen in gastrulation, where cells segregate from ectoderm, gradually cease to express E-cadherin and begin to express N-cadherin [25]. Noteworthy, in early stage of embryonic morphogenesis of mouse submandibular gland, N- and E-cadherin display diffuse expression at cell-cell contacts, but at late stage of postnatal development, N-cadherin disappears while E-cadherin expression progressively increases [26]. Perhaps, N-cadherin expression in this case of Warthin’s tumor indicates that the tumor is still in a developmental stage in contrast to the 9 cases with completed development. Further investigations will be necessary to confirm our finding and possibly explain the role of N-cadherin in Warthin’s tumor. In many cases of Warthin’s tumors, N-cadherin expression was observed, for the first time, in membranes of cells in germinal centers of lymphoid tissue. Centrocytes have the ability to migrate [27] and N-cadherin expression is possibly essential for this movement and contacts formation between them and follicular dendritic cells that originate from mesenchyme [28].

The immunohistochemical examination of twenty-nine malignant salivary gland tumors showed N-cadherin expression in 13 (45%) tumors. The N-cadherin expression was not significantly correlated with the stage of tumors including mucoepidermoid carcinoma as previously reported by Shieh et al. [20] who studied N-cadherin expression in mucoepidermoid carcinomas. However, they evaluated the presence of N-cadherin in mucoepidermoid carcinomas in comparison with the positive staining of N-cadherin in normal salivary gland tissue, as well. In contrast, our results revealed a de novo expression of N-cadherin in tumor cells in comparison to the absent expression of this molecule in the normal salivary gland tissues. This is probably due to a switching mechanism including the reduction of E-cadherin and the increase of N-cadherin expression during neoplastic process in salivary gland neoplasms as observed in several types of cancer [29,30]. This reduced E-cadherin expression has been detected in cases of ACC, MEC, and PLGAs in our previous study [23].

E- to N-cadherin switching usually refers to the loss of E-cadherin expression and the gain of N-cadherin presence but may also include situations in which E-cadherin expression levels do not change significantly but the cells turn on a synchronous expression of N-cadherin, or other cadherins. Cells that express significant amounts of E-cadherin but only a small amount of N-cadherin may have increased motility [9].

In the present study, N-cadherin was almost totally absent in benign neoplasms, but was expressed almost in half of the malignant ones. This pattern of expression, together with the significant correlation between perineural invasion and N-cadherin expression in malignancies suggests a role for N-cadherin as an indicator (biomarker) of potentially aggressive biologic behavior of malignant salivary gland tumors in terms of neoplastic cell motility and invasion enhancement especially via neural tissue. Interestingly, N-cadherin-bearing tumor cells gain motility, form transient contacts with the surrounding extracellular matrix, approach, interact and invade neural tissue [11] as well as vascular endothelium [15] that express N-cadherin also.

N-cadherin expression in malignant salivary gland tumors may be an indicator of potentially aggressive behavior. Perineural invasion likely is attributed to an N-cadherin expression by tumor cells.

Mr. N. Kehagias was supported financially by the Program ‘Heraklitos II’.