SCSGs are generally low-grade salivary gland carcinomas, and are characterized by morphological resemblance to mammary analogue secretory carcinoma and ETV6–NTRK3 gene fusion [2, 7]. In 2017, SCSGs were added to the WHO classification of head and neck tumors [3].
The age of onset of SCSGs reportedly varies widely. These tumors usually develop in adults, but can occur in children and adolescents; there is no significant sex difference [8]. In the present group, the median age was 45 years (range: 10–69 years) (male:female ratio 1.3:1). The parotid gland is the most common site of SCSGs; small salivary glands such as the buccal, upper lip, and palatal glands can also be involved [9, 10, 11]. The tumors of 21 of the 23 patients in the present study were located in the parotid gland, 17 being in the right parotid and four in the left parotid gland, differing slightly from that previously reported [12]. SCSGs most commonly present as painless, slow-growing, well-circumscribed masses, and can be misdiagnosed as pleomorphic adenoma or acinar cell carcinoma [13]. Consistent with previous reports, most patients presented after finding a painless mass, with only a few recurrences being sensitive to touch. In the present study, no specific pre-operative ultrasonography, CT or MRI findings were identified. The tumors presented as well-demarcated masses of uneven density, sometimes with cystic changes, that could easily be misdiagnosed as pleomorphic adenomas. There were only three patients with T3 tumors in our cohort, the others being T1 or T2 tumors. However, T4 tumors have been reported by others [12].
Before their recognition, SCSGs were frequently classified as ACCs because of the nearly identical histological growth patterns of these tumor types [14]. There are no significant clinical differences between SCSGs and ACCs. However, these tumors may show considerable morphological differences histopathologically. Unlike ACCs, SCSGs have no secretory zymogen cytoplasmic granules that show true positive periodic acid–Schiff reactions [15]. The nuclei of SCSGs have prominent pseudo-inclusion bodies and SCSGs do not have the other cell types, such as serous, intermediate, and clear cells, that are seen in ACCs [16]. As to immunophenotypes, SCSGs can express various markers of breast secretory carcinomas, including mammaglobin, S-100 protein, STAT5A, MUC1, MUC4, GCDFP-15, as well as CK7, CK8, CK18, CK19, and epithelial cell membrane proteins [17]. However, they characteristically do not express DOG-1, estrogen, androgen, progesterone receptors, or HER2 [18]. In contrast with SCSGs, ACCs do not express mammaglobin, or express it weakly locally, whereas they express DOG-1 diffusely [18]. Khalele et al. showed that SCSG can be diagnosed when a salivary gland tumor expresses mammaglobin and S-100 protein strongly, and does not express DOG-1 [19]. It has also been reported that if mammaglobin and S-100 are strongly positive, SCSG can be diagnosed without resorting to molecular biological methods [20]. In particular, mammaglobin has become a very important tumor marker for diagnosing SCSGs.
Similarly to secretory carcinomas of the breast, SCSGs have been shown to have the translocation t(12; 15) (p13; q25), which results in fusion of the ETV6 gene on chromosome 12 with the NTRK3 gene on chromosome 15 [4]. ETV6–NTRK3 gene fusion has not been found in other types of salivary gland tumor. However, it has been found in congenital fibrosarcoma, mesodermal renal tumor, and acute myeloid leukemia [21, 22]. Recent findings have expanded the molecular profile of SCSG to include multiple novel ETV6 fusion partners, including ETV6-RET and EGFR–SEPT14 [23]. Rupture of ETV6 and its fusion with NTRK3 may lead to production of functional tyrosine kinases or affect the signaling pathways of some transcriptional activators, leading to strong expression of mammaglobin and S100, but not of DOG-1 [24]. It is therefore considered that detection of ETV6 gene rearrangement is the most specific method, and therefore the gold standard, for diagnosing SCSGs. Additionally, gene therapy may offer new hope to patients with this tumor in the future [25]. In our study, the tumors of 15 patients that had not been definitely diagnosed by immunohistochemistry were diagnosed on the basis of ETV6 gene rearrangement, this being found in all 15 cases. Chiosea reviewed 81 cases of ACC for ETV6 fluorescence by in situ hybridization and found that all classical ACCs had intact ETV6, whereas 10 of 17 zymogen granule-poor ACCs showed ETV6 translocation; these were accordingly reclassified as mammary analog secretory carcinomas [26]. Other differential diagnoses include adenoid cystic carcinoma, mucoepidermoid carcinoma, and salivary ductal carcinoma, all of which can identified by immunohistochemistry and ETV6 gene detection. Before immunohistochemistry and ETV6 gene detection had resulted in the correct diagnoses of SCSG being made, four of the present 23 patients’ tumors were misdiagnosed as acinar cell carcinomas, three as mucoepidermoid carcinomas, two as adenoid cystic carcinomas, one as low-grade ductal carcinoma, and one as squamous cell carcinoma. Thus, the rate of misdiagnosis with routine pathology only was 47.8%.
Because SCSGs behave similarly to ACCs, the treatment of SCSGs is mainly surgical. However, there is currently no consensus on optimal treatment of these tumors. Some researchers believe that SCSGs are indolent tumors that rarely produce distant metastases and that surgery alone is sufficient [27]. There have also been reports that surgery combined with postoperative radiotherapy is more effective [28]. In the present study, 11 patients were treated with surgery alone, one of whom had a recurrence postoperatively. Radioactive I125 implantation was performed under MRI navigation in four patients, none of whom developed recurrences. In addition, one patient received postoperative chemotherapy and later had a recurrence. Four of seven patients who received postoperative chemoradiotherapy had recurrences. The overall recurrence rate was 26.1%.
Interestingly, six of the patients with recurrences had undergone their first surgeries in local primary hospitals. Their recurrences were considered to have resulted from incomplete resection by the first surgery. There are currently few reports of different surgical procedures for SCSGs. Some studies advocate simultaneous total resection of the affected lobe of the parotid gland and cervical lymph node dissection [29]. Chiosea et al. reported a higher incidence of lymph node metastases from SCSGs than from ACCs (33% vs. 8%); despite this, cervical dissection is still rarely performed [30]. Five of the present 23 patients developed cervical lymphatic metastases and two developed distant metastases (both to the lung) after having local recurrences. None of the patients who had undergone excision of the tumor and superficial lobe of the parotid gland had recurrences or distant metastases. All six patients with local recurrences had only undergone tumor enucleation or partial parotid gland resection, which suggests that preoperative determination of the size and extent of the tumor is vital to the success of surgery.
It has recently been shown that MRH of salivary gland ducts can clearly establish the three-dimensional relationship between the tumor and parotid duct and detect dilation, stenosis, displacement, and destruction of main ducts and branch ducts, potentially guiding preoperative planning [31]. If the tumor has invaded the main duct, retaining the gland is pointless and superficial or total parotid gland resection should be performed where possible. If the tumor is sufficiently distant from the main duct, regional parotid gland resection with preservation of part of the gland can be considered. Thus, detailed MRHs can accurately guide the planning of parotid gland preservation surgery, potentially enabling retention of some functional gland while completely resecting the tumor. In the present study, two patients had undergone preoperative magnetic resonance hydrography and regional resection of the parotid gland without any other postoperative treatment. Thus far, they remain well, their average follow-up time being 26 months.
The prognosis of SCSG is related not only to the first operative procedure, but also to age, clinical stage, and Ki-67 proliferation index. Although six of 23 patients in this study had local recurrences (mean age 53 years) and one patient died, the overall prognosis was good, which is consistent with reported results [32]. Additionally, the prognosis is age-dependent. Children and young patients have better prognoses, adult patients having more aggressive tumors. Older patients have the most aggressive tumors and highest recurrence rate [33]. This may be related to differences in pathological features at different ages and the characteristically poor immune function of older patients. Furthermore, previous studies have shown that high clinical stage and high grade are the main adverse prognostic factors, lymph node involvement and extra-parenchymal glandular invasion being associated with a greater risk of local recurrence and metastasis [34]. In the present study, five patients developed lymph node metastases, all of whom underwent therapeutic neck dissection and postoperative radiotherapy and/or chemotherapy. However, they all had postoperative recurrences, with pulmonary metastases in two of them, one of whom died. Additionally, the tumor of the patient who died had a Ki-67 proliferation index of 50%+ and had recurred three times. This is consistent with reports that patients with a Ki-67 proliferation index of more than 10% have poor prognoses, suggesting that the Ki-67 proliferation index affects the prognosis of secretory carcinomas [35].