Salivary Epithelial-Myoepithelial Carcinoma: Clinical, Histopathological and Molecular Study from A Latin American Case Series with a Novel MSH3 Driver

Significance

And Novelty

Statistical Analysis Declaration

The present study aims to analyze samples of epithelial myoepithelial carcinoma of salivary glands (EMC) collected from Brazilian patients and perform clinicopathological correlations. For the first time in the literature, the analysis of a Latin American population with EMC is reported. Moreover, an exploratory analysis of the tumor mutation burden was performed.

Statistical methods employed are performed by an experienced person, with authorship on the manuscript, and the contributor specified.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12105-025-01776-0.

Abstract

Purpose

Epithelial-myoepithelial carcinoma (EMC) is a rare malignant neoplasm of the salivary glands. Although most studies have reported the predominant, recurring driver mutation is in HRAS, its molecular pathogenesis is not yet fully understood. The aim of this study is to describe the clinical and histopathological characteristics of a case series of salivary EMC with non-RAS genetic variants.

Methods

Fifteen formalin-fixed paraffin-embedded (FFPE) samples of EMC were retrospectively retrieved. Clinicopathological data was recovered from medical records. Two cases with sufficient tissue were assessed through next-generation sequencing (NGS).

Results

EMC occurred more frequently in females (60%). The mean age at diagnosis was 59.5 (± 14.5) years. The parotid was the most common primary site (80%). Time from first symptoms to diagnosis was 41 months in average. Pathological staging I, II and III occurred in 21.4%, 28.6% and 50% of the cases, respectively. Only one participant had regional lymph node involvement, and none were metastatic. The margins were involved in 41.6% of cases. All the patients were surgically treated, including neck dissection in two subjects, and 33% received adjuvant radiotherapy. One patient relapsed (6.7%) after an average of 32.4 months of follow-up. NGS analysis revealed 160 germline mutations in 46 genes of interest, of which 11 were previously described as pathogenic or as variant of uncertain significance (VUS). Only one somatic frameshift deletion involving MSH3 (c.1148del p.K383Rfs*32) gene was detected.

Conclusion

This study explores EMC in a Latin American cohort. The findings align with global literature, and a novel MSH3 mutation was identified. However, further research is needed to confirm its significance and potential impact on tumor behavior.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12105-025-01776-0.

Introduction

Epithelial-myoepithelial carcinoma (EMC) represents 1 to 2% of salivary gland tumors [1, 2]. It is defined as a malignant tumor of the salivary gland with biphasic tubular structures. EMC is composed of variable proportions of two cell types: an inner layer of duct-lining cells and an outer layer of clear cells, which typically form double-layered duct-like structures [3]. In general, EMC are low-grade tumors, affect mainly patients between the fifth and eighth decades of life, predominantly in women and caucasians [4, 5]. Most EMC originate in the parotid gland (57.7%), followed by the submandibular gland (9.8%) [6].

The diversity of EMC subtypes and clear cell morphology expands the differential diagnosis, encompassing other salivary gland tumors with biphasic and/or clear cell morphology, primarily benign entities such as pleomorphic adenoma and basal cell adenoma. However, malignant salivary gland tumors including adenoid cystic carcinoma, basal cell adenocarcinoma, myoepithelial carcinoma, and clear cell carcinoma, should also be considered [1, 6, 7]. Recently, HRAS gene mutations were described in EMC [8–12] As the HRAS mutation has not been described in other salivary gland tumors with similar morphology, some authors have suggested that the HRAS mutation test may be useful for the diagnosis of this neoplasm [13].

Here, we aimed to study the clinical and histopathological characteristics of a case series of salivary EMC from a Latin America population and to investigate new genes of interest in the tumor.

Materials and Methods

Ethics Statement

This study was approved by the Ethics Committee of the Faculty of Medical Sciences of Minas Gerais, Belo Horizonte, Brazil (CAAE: 6.578.178). All procedures were in accordance with the ethical standards of the committees responsible for human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008.

Study Design

It is a cross-sectional study carried out on individuals over 18 years old with a histopathological diagnosis of EMC of the major and minor salivary glands.

Sample Selection

Due to the low incidence, a convenience sampling was carried out. All the samples were retrieved from the Surgical Pathology Laboratory of the University of Campinas, Federal University of Bahia, and Federal University of Minas Gerais. Of all the cases from these institutions, there were 4,387 malignant head and neck tumor cases, 157 of which were malignant salivary gland tumors, and 25 were diagnosed as EMC based on the essential criteria proposed by the WHO classification. However, only cases with adequate clinicopathological descriptions in the medical records were included, resulting in a final sample of 15 EMC cases in this study For each case, clinical information was collected: sex, age, location of the primary tumor, tumor evolution, recurrence, tumor size, staging, status of neck dissection, perineural and vascular invasion, margin involvement, metastasis, treatment, status of the patient (alive without evidence of disease, alive with active disease or deceased), and follow-up time.

Of the fifteen EMC, FFPE blocks were available for additional testing on two cases. Two experienced pathologists (R.S.G and R.R.M.C) reviewed the H&E and immunohistochemical slides. The diagnosis of EMC was based on the following histopathological criteria: low-grade tumor, with multilobular invasive growth, biphasic morphology, with a tubular structure of internal eosinophilic ductal epithelial cells (luminal) and external clear myoepithelial cells (abluminal) (Fig. 1). All selected cases showed a well-defined stromal layer, delineated by a prominent basement membrane, which contributed to the differential diagnosis with pleomorphic adenoma. Additionally, the myoepithelial component did not exhibit hyperchromatic and angular cells, which are characteristic of adenoid cystic carcinoma. Immunohistochemistry should show positive staining for AE1/AE3, CK7 and/or EMA in the ductal cells and positivity for p40, smooth muscle actin, S100 and/or calponin in the myoepithelial cells (Fig. 2). No cases of ex-pleomorphic adenoma were noted.

Fig. 1.

ECM histopathology. (a) A low-power view of the tumor surrounded by a capsule. (b) and (c) Histopathological features of the two sequenced cases (#9 and #10). It presents biphasic tubular structures, composed of tightly coupled internal ductal cells and clear myoepithelial cells arranged externally (original magnification, H&E staining, x20)

Fig. 2.

Classic ECM immunohistochemistry. (a) Immunohistochemistry to show ductal cells. CK7, demonstrating strong positivity in ductal cells. (b) Immunohistochemistry to evaluate myoepithelial cells. Calponin was used, which stains the cytoplasm of myoepithelial cells. (c) Staining with p40, which stains the nucleus of myoepithelial cells

DNA Extraction and next-generation Sequencing

Genomic DNA (gDNA) was isolated using the QIAamp DNA FFPE Tissue Kit (QIAGEN, Hilden, Germany) according to manufacturer’s instructions. DNA was quantified using a spectrophotometer (Nano-DropTM 2000 instrument) [14]. Libraries were prepared using the pre-made QIAseq^® Targeted DNA Tumor Mutational Burden Panel kit (QIAGEN, Hilden, Germany). After library construction and purification, quality control was performed with TapeStation [15] to verify the correct size distribution of library fragments and the absence of primer. Subsequently, samples underwent polymerase chain reaction (PCR) to amplify the target regions.

Sequencing using Illumina technology was then carried out using the NextSeq 550 equipment, using 2 sequencing kits of the NextSeq 550 System Mid-Output Kit type (Illumina, San Diego, USA) with 300 cycles. Sequencing was paired-end, with readings of 8 bases from each index (151 × 8 × 8 × 151).

Statistical Analysis

A descriptive statistical analysis was performed. Mean, median and standard deviation were used for quantitative data. Relative frequencies and absolute values were used ​​for nominal variables.

NGS data were analyzed using the CLC Genomics Workbench (QIAGEN), by the workflow ‘Identify QIAseq DNA Somatic Variants’ (Illumina) available at the plugin ‘Biomedical Genomics Analysis’. Clinvar [16], COSMIC [17], and Gnomad [18] were used to retrieve information regarding the detected mutations.

Results

Sample Characterization

Samples from 15 EMC tumors were included. Nine participants were female (60%) and 6 male participants (40%). The mean age at diagnosis was 59.3 years, median 65 ± 14.4 years (Table 1).

Table 1.

Clinical characteristics of patients with EMC

Case	Age (years)	Sex	Primary tumor	Time between initial symptoms and diagnosis (months)	pT(cm)	pN	Pathological initial staging
1	50	F	Parotid	24	8	1	III
2	30	F	Parotid	24	7.5	0	III
3	66	F	Parotid	NI	1.5	0	I
4	64	M	Parotid	240	4	0	III
5	66	M	Parotid	18	4	0	III
6	59	F	Parotid	12	4	0	III
7	51	M	Palate	NI	3	0	II
8	84	M	Palate	8	2.2	0	II
9	65	M	Parotid	12	3	0	II
10	52	F	Parotid	24	1.5	0	I
11	32	F	Floor of mouth	8	8	0	III
12	67	F	Parotid	NI	NI	NI	NI
13	71	F	Parotid	NI	0.7	0	I
14	71	M	Parotid	NI	5	0	III
15	65	F	Parotid	NI	3	0	II

F: female, M: male, NI: not informed, pT: pathological staging (tumor size). pN: pathological staging (lymph node involvement. Staging based on 8th American Joint Committee on Cancer (AJCC) edition [19]

Regarding the primary site of the lesion, 12 occurred in the parotid (80%), two affected the minor salivary glands of palate (13.3%), and one developed in the floor of mouth (6.7%). Time from first symptoms to diagnosis was available for 9 participants. The average evolution time was 41 months (16 ± 74.8), ranging from 8 to 240 months. Local swelling was present in all 12 cases in which symptom information was available. Pain was present in three cases (25%), one had local bleeding (8%), one (8%) had weight loss and one (8%) had difficulty rotating their neck.

Pathological staging was performed in 14 cases. The average tumor size was 3.8 cm (5 ± 2.18). Pathological staging I, II and III occurred in 21.4%, 28.6% and 50% of cases, respectively. Only one participant (#1) had regional lymph node involvement at diagnosis, and none were metastatic (Table 1). A positive surgical margin was observed in 5 of 12 cases. Perineural invasion was present in only one case and no cases had vascular invasion (Table 2).

Table 2.

Treatment and follow-up information

Sample	Treatment	Cervical dissection		Vascular /Perineural invasion		Margin	Recurrence			Follow- up time
1	Surgery	Yes	No		Positive			No	NI
2	Surgery	No	No		Free			No	NI
3	Surgery	No	No		Positive			No	NI
4	Surgery + RDT	Yes	Yes		Free			No	45*
5	Surgery + RDT	No	No		NI			No	72
6	Surgery + RDT	No	No		Positive			No	18
7	Surgery	No	No		NI			No	12
8	Surgery	No	No		Free			No	12
9	Surgery + RDT	No	No		Positive			No	74
10	Surgery	No	No		NI			No	20
11	Surgery + RDT	No	No		Free			No	6
12	Surgery	No	NI		Positive			Yes	NI
13	Surgery	No	No		Free			No	NI
14	Surgery	No	No		Free			No	NI
15	Surgery	NI	No		Free			No	NI

RDT: radiotherapy. NI: not informed. *Deceased from a secondary cause

All included patients had surgical excision, and two had concomitant neck dissection due to the regional lymph node involvement in patient #1 and the vascular/perineural invasion observed in the initial biopsy specimen of patient #4. Adjuvant radiotherapy was performed in 5 patients (33%). Of the patients who received adjuvant radiotherapy, one had perineural invasion (7%), two had compromised margins, one was stage III with a large lesion (8 cm), and one was stage III with a non-evaluable margin. No patient was treated with chemotherapy. Only one patient had a recurrence (6.7%), which occurred 17 months after diagnosis. This recurrent tumor was excised, and the patient was submitted to radiotherapy.

Information on clinical follow-up was available in 9/15 participants. One died from a secondary cause and the remaining were alive with no evidence of disease (88.8%). The average follow-up time was 32.4 months (19 ± 27.6).

Next-generation Sequencing

Of the fifteen cases, we were able to retrieve FFPE blocks for only two. One sample harbored 1,166 mutations and 1251 variants were detected in the other. Ninety genes of interest for salivary gland neoplasms were selected, based on a literature review [8–11, 13, 20–25]. Genes of interest were those in which mutations had already been described in salivary gland tumors.

To identify germline variants, a reference dataset from the Single Nucleotide Polymorphism Database (dbSNP), maintained by the National Center for Biotechnology Information (NCBI), was used. Altogether, 160 germline mutations were detected in 46 genes of interest, of which 11 had at least one description in the literature as pathogenic mutation or as variant of uncertain significance [16–18]. Fifty-three mutations without previous records in these databases were identified. The other 65 mutations had already been described as benign or probably benign. No HRAS p.Q61R variant was detected in the two sequenced cases. The other thirteen cases were not assessed for HRAS gene mutations. The mutations characterization is depicted on Supplementary Material 1. Only one somatic mutation was detected. It was a deletion involving the MSH3 gene (c.1148del; p.K383Rfs*32– COSM1438888) previously described in others malignant neoplasms, such as malignant melanoma [26–28], adenocarcinoma and metaplastic carcinoma. However, despite its association with other malignancies, no history of benign or malignant neoplasms was identified in our patient.

Discussion

To our knowledge, this is the first clinicopathological study on EMC from a Latin American population. The clinicopathological tumor findings align with the existing literature. There was a predominance of primary parotid tumors, and it was prevalent among women and individuals between the fifth and eighth decade of life. The lymph node involvement rate was low (6%) and none of the patients exhibited local or distant metastasis. These features are similar to those described by Gore [6], who analyzed the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2014.

The median time between symptom onset and diagnosis was 41 months. This reflects the slow evolution of the tumor and symptoms with little impact on patients’ quality of life. The median tumor size was 3.5 cm, similarly to that described by other authors [1, 13]. The reported recurrence rate for salivary gland EMC is about 40% [5, 29, 30]. In the present study, one case was a recurrent EMC. This case was the only one in which the researchers did not have access to material from the primary tumor. Due to this, information about initial characteristics and risk factors could not be analyzed. The lower recurrence rate in our cohort, compared to the literature, may be related to the follow-up time. The median follow-up was only 19 months, a short period for this type of tumor.

The tumor mutation burden panel revealed a somatic mutation on MSH3 gene that is seemingly the driver mutation in this case, and to our knowledge, the first report of this driver for EMC. The MSH3 gene produces a protein that is involved in the DNA mismatch repair system. Loss of function of the MSH3 protein leads to microsatellite instability and increased risk of cancer [31]. Amaral-Silva and colleagues carried out a study that evaluated the expression of MSH3 in salivary gland tumors. It was shown that MSH3 expression was lower in malignant salivary gland tumors compared to benign tumors and normal salivary gland tissue [32]. They also found that adenoid cystic carcinoma with perineural invasion also showed lower MSH3 expression, indicating that loss of MSH3 protein expression is related to more aggressive phenotypes (malignancy and perineural invasion) [32].

Tumors with a defective mismatch repair system and microsatellite instability have shown a favorable response to immune checkpoint inhibitors, such as anti-PD-1 antibodies [33]. Although most EMC are cured with surgical resection, some cases may undergo high-grade transformation, which is a poor prognosticator [34]. Considering the functional impact of MSH3 defects and their potential role in genomic instability, further investigation into this target in EMC may provide insights into the feasibility of immunotherapy-based treatments for recurrent or metastatic disease. In our case with mutated MSH3, there was no associated high-grade transformation.

Although mutations affecting the codon 61 of HRAS gene have been described as EMC hallmarks, neither of our assessed cases had this mutation. The frequency of HRAS mutation varies from 26.7 to 81.7% [13, 23]. It is important to highlight that only two samples were evaluated and that the non-identification of the mutation does not exclude its relevance for EMC.

The study has the limitation of having only 2 samples included in the molecular analysis. While the broader molecular approach used is promising to find novel EMC driver mutations, additional analysis are required to further validate our findings.

Conclusion

While this study provides a case series of EMC in a Latin American population, the molecular assessment was limited to only two cases. The clinicopathological characteristics corroborate with global literature. Although a seemingly novel oncogenic driver mutation in MSH3 was reported, further investigation is crucial to validate this finding and assess its functional impact on tumor behavior.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author Contributions

M.S.F. wrote the main manuscript text and P.V.S and M.H.A prepared Figs. 1 and 2. R.G.M. performed the statistical analysis. F.V.M.C. and J.N.S. contributed with tumor samples. M.G.D and R.R.M.C. performed the laboratorial experiments. R.R.M.C. and R.S.G. conceptualized the research and critically revised the manuscript. All authors reviewed the manuscript.

Funding

The study had FAPEMIG funding.

Data Availability

Data can be obtain by contacting corresponding author.

Declarations

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent to Participate

This study has obtained IRB approval from (Ethics Committee of the Faculdade de Ciências Médicas de Minas Gerais) and the need for informed consent was waived.

Consent to Publication

For this type of study consent for publication is not required.

Competing Interests

The authors declare no competing interests.

Code Availability

No software application or custom code was used in the preparation of this article.