Introduction

Salivary gland epithelial-myoepithelial carcinoma (EMC) is a low-grade tumor characterized by a typical biphasic epithelial-myoepithelial growth pattern, clear myoepithelial differentiated abluminal cells and intraluminal secretions [1, 2]. EMC is in most cases lobulated, well circumscribed and/or encapsulated mimicking a benign lesion including (cellular) pleomorphic adenoma, whereas EMC ex pleomorphic adenoma (ex PA) can also be observed [1, 3]. Recently, morphological variants such as oncocytic (-sebaceous), apocrine and double-clear have been described, extending the morphological spectrum of EMC [4, 5]. Rarely, high-grade transformation, also known as “dedifferentiation”, can be observed in a minority of EMC cases, characterized by necrosis, anaplasia and/or myoepithelial overgrowth [3]. On a molecular level, recurrent HRAS mutations can be found in EMC, and PLAG1/HMGA2 translocations in EMC ex PA [3, 6]. Here, we describe an unusual case of an overwhelmingly solid-oncocytic differentiated EMC in the parotid gland of a 70-year-old female patient, with two tumor clones, both harboring a comprehensive HRAS p.Q61R mutation and a heterozygous loss of ARID1A, while only one of the clones presented with a hitherto, in this entity unpublished, ARID1A terminating mutation.

Case Report

Clinical Presentation

A 70-year-old female was referred to our Department of Otorhinolaryngology—Head and Neck Surgery for a painless nodule at the angle of the left mandible for a few months. Clinical exam revealed a mobile firm mass in the deep parotid. Facial nerve function was symmetrical. Ultrasound showed a hypoechoic well-circumscribed round mass of up to 1.95 cm in diameter inside the parotid gland, directly posterior to the mandible. No enlarged or suspicious lymph nodes were detected by MRI and ultrasound of the neck.

Cytology

The ultrasound-guided fine needle aspiration showed a cellular salivary gland neoplasm with an epithelial/myoepithelial differentiation and discreet atypia. Differential diagnosis encompassed basal cell adenoma, cellular pleomorphic adenoma and low-grade salivary gland carcinoma including epithelial-myoepithelial carcinoma.

Resection Specimen Morphology

The nodule was arising from the deep parotid lobe. It was resected completely without sacrifice/injury to any branches of the facial nerve. The postoperative period was uneventful showing symmetrical facial nerve function. The resection specimen was submitted entirely and formalin-fixed, paraffin-embedded. Tissue was examined on 2 µm-thick hematoxylin & eosin (H&E) stained sections. The H&E morphology showed a very well circumscribed and encapsulated, lobulated cellular salivary gland neoplasm (Fig. 1a) with sectional obvious biphasic epithelial-myoepithelial differentiation, including minimal cleared myoepithelial cells, intraluminal secretions and some calcifications (Fig. 1b). Focal prominent fibrous stroma was noted (Fig. 1c). Besides, overwhelming solid-oncocytic differentiation was evident, in which a biphasic growth pattern was more difficult to appreciate (Fig. 1d). Some mitotic figures were visible. Focal infiltration into the capsule and intravascular tumor complexes could be noted (Fig. 1e), but no convincing extension beyond the capsule. No perineural infiltration or areas of a pleomorphic adenoma were apparent.

Fig. 1
figure 1

Overview of this epithelial-myoepithelial carcinoma. a Shows the lobulated and encapsulated cellular neoplasm. In b a more typical epithelial-myoepithelial differentiation is seen with cystic epithelial structures filled with secretions and focal calcifications. c Depicts the focal prominent fibrous stroma dissecting the epithelial cells, whereas d shows the more solid-oncocytic differentiation, corroborated by p63 and CD117 staining (insets) visualizing the biphasic, epithelial-myoepithelial differentiation. In e intravascular tumor complexes are visible in the capsule region, being surrounded by endothelium (different slide as shown in (a). Scale bar 2.5 mm (a) and 250 μm (b–e)

Full size image

Resection Specimen Immunohistochemistry

The overall biphasic population of epithelial and myoepithelial cells could be corroborated by positive staining for p63 (abluminal myoepithelial cells) and CD117 (luminal epithelial cells) in both compartments (Fig. 1d insets, Table 1). Virtually all tumor cells showed diffuse nuclear positivity for SOX10 in both cell types (epithelial and myoepithelial cells), and all areas. No nuclear expression of PLAG1 or β-Catenin was present. MIB-1 proliferation index was worrisome at around 10–15%. A typical wild-type pattern of p53 staining was found. Staining for ARID1A protein showed two distinct tumor clones, whereas the expression was especially lost in the solid-oncocytic differentiated areas and preserved in the more classical, tubular areas (Fig. 2).

Table 1 Overview of the immunohistochemical stainings performed
Full size table
Fig. 2
figure 2

Morphological overview of different tumor clones in this epithelial-myoepithelial carcinoma. a Shows more classical region with cystic glandular structures containing prominent secretions in the lower right corner. b Shows preserved expression of ARID1A in this region, whereas the adjacent solid-oncocytic differentiation demonstrates total loss of ARID1A expression. In c a magnification of the separating line between the two clones can be appreciated. Similarly, d shows another region with lightening morphology in a central and right beside nodule, as well as more tubular differentiation in the left upper corner. These nodules show strong expression of ARID1A (e), whereas the adjacent darker regions have lost the expression. In f a magnification of the two tumor clones can be seen with prominent secretions in the ARID1A conserved clone on the left. Scale bar 1 mm (a–b, d–e); scale bar 100 μm (c, f)

Full size image

Molecular Testing Using Next Generation Sequencing

Next generation sequencing (NGS) using the Oncomine comprehensive assay v3 (LifeTechnologies) revealed a HRAS p.Q61R mutation in both investigated areas. Furthermore, both areas had a heterozygous loss of ARID1A. Additionally, a terminating ARID1A mutation (p. E672*) was found in the solid differentiated areas while ARID1A was wildtype in a more classical tubular area, corresponding to loss of expression and expression in the ARID1A immunohistochemistry, respectively. No fusions of PLAG1- or HMGA2 were found using the Archer FusionPlex Sarcoma panel (ArcherDx).

Finally, diagnosis of low-grade EMC with solid-oncocytic differentiation was provided, corroborated by the typical described HRAS p.Q61R mutation. Interestingly, the terminating ARID1A mutation was found only in one tumor clone, whereas the other clone was ARID1A wildtype in the molecular testing. These results correlated well with the expression pattern of ARID1A protein by immunohistochemistry.

The tumor was resected completely with negative surgical margins. The histology was discussed at the local interdisciplinary head & neck tumor board with decision that no adjuvant treatment was required. No follow up data are available because the case was recent.

Discussion

We report the first case of a heterogeneous ARID1A mutation with loss of heterozygosity in addition to a typical HRAS p.Q61R mutation in a salivary gland EMC. So far, HRAS mutations were found in around 30% of EMC in the literature [3, 6], whereas a very recent study reported HRAS mutations in up to 82.7% of EMC in a large cohort including different morphologic subtypes [7]. Furthermore, cases with isolated PLAG1 or HMGA2 translocations have been described, typically encountered in cases with EMC ex PA [3]. Rarely, EMC shows high-grade transformation/dedifferentiation, whereas in a larger study by El Hallani et al. additional aberrations in TP53, FBXW7 and SMARCB1 were found [3]. In breast adenomyoepitheliomas, which are morphologically related to EMC, similar findings with HMGA2 translocations, activating HRAS mutations or mutations in PIK3CA and AKT1 were described recently [8, 9]. The AT-rich interactive domain-containing protein 1A (ARID1A) encoded by the ARID1A gene is part of the SWI/SNF chromatin remodeling complex and has a tumor-suppressing function [10]. ARID1A mutations have been very well characterized in ovarian clear cell carcinoma and endometrioid adenocarcinoma [11], whereas aberrations in ARID1A have been described only very rarely in salivary gland tumors: a ARID1A-PRKD1 fusion has been reported in cribriform adenocarcinoma of salivary gland origin by Weinreb et al. [12]. Moreover, Sebastiao et al. recently showed a terminating p.Q288Pfs*71 frameshift ARID1A deletion in a polymorphous adenocarcinoma [13]. Terminating mutations and frameshift mutations leading to early termination of ARID1A protein translation and thereby to a loss of function, have been described previously in salivary gland carcinomas (12/114 sequenced salivary gland tumors) encompassing in addition salivary duct carcinoma (2/18 samples) and adenoid cystic carcinoma (10/73 samples) [14]. However, none of these samples showed a heterozygous deletion of ARID1A in addition. ARID1A mutations are known in various types of cancer, whereas they can affect either one or both alleles, e.g. in ovarian clear-cell carcinomas [15]. While some of the tumors retain protein expression, most of them are lacking it, suggesting haploinsufficiency of ARID1A. The EMC of our patient showed two different morphologies. While the HRAS mutation was most likely the driver in both of them, loss of heterozygosity of ARID1A tumor suppressor acted as a probable second hit and was predominantly found in the more solid-oncocytic than the more classical tubular growth pattern. In vitro and in vivo studies had shown anti-tumor activities of several targeted inhibitors, such as ATR-, PARP- or AKT inhibitors [16,17,18]. These findings have recently led to clinical trials that investigate the effectiveness of several targeted monotherapies or combinational therapies, including combinations with immunotherapy (NCT03718091, NCT02286687, NCT03682289, NCT02576444, NCT03842228, NCT03297424), in solid tumors.

Conclusion

Epithelial-myoepithelial carcinoma with solid-oncocytic differentiation may harbor besides HRAS mutations additional ARID1A terminating mutations. The loss of ARID1A expression in one tumor clone may be a “second hit” and suggests a morpho-molecular association, as it was mainly found in the solid-oncocytic differentiation. Additional studies of this rare entity are needed to further assess these considerations.