Introduction
Malignancies of the salivary glands are rare and account for approximately 5–8.5% of all head and neck cancers (HNC) [1–3]. Their occurrence is rare, with an annual incidence of 0.69 cases per 100,000 [4, 5]; however, the mortality rate is 40% [1]. Moreover, an increase of approximately 50% in both morbidity and mortality is predicted in the near future [6]. Salivary gland cancers (SGCs) are characterised by miscellaneous disease courses and clinical behaviours that contribute to unfavourable patient outcomes [2]. Among SGCs, more than 20 histopathological varieties have been classified by the World Health Organisation. Mucoepidermoid carcinoma (MEC) is the most common type of cancer, followed by acinic cell carcinoma (AcCC), adenoid cystic carcinoma (AdCC), carcinoma ex-pleomorphic adenoma (Ca ex PA), and adenocarcinoma (AC) [2, 7, 8]. The number of histopathological features interfering with benign lesions might also contribute to misdiagnosis and inappropriate management [9–11]. The incidence of these tumours is greater in males, and the risk of development increases with age. Former exposure to radiotherapy is also a well-known risk factor [3, 12–15]. A history of other cancers, including HNC, and occupational hazards are also associated with SGC incidence [3, 13]. In contrast to HNC risk factors, neither alcohol consumption nor tobacco use increases the risk of salivary gland malignancies [12, 13]. Numerous other causative factors have been proposed; how-ever, studies are limited, and the results are inconclusive. Suspicious lesions, especially those with rapid growth, associated painful swelling, facial nerve palsy, or ulceration, indicate malignancy and should be investigated by imaging methods, preferably multiparametric magnetic resonance imaging.
Preoperative fine-needle aspiration enables the differentiation between benign and malignant tumours as well [2, 3, 16]. Radical surgical excision is the standard management option. Owing to tumour advancement and histopathological features, patients must receive further adjuvant radiotherapy or chemoradiotherapy [5, 16]. Park et al. reported disease recurrence in more than 50% of SGCs, despite radical primary treatment [17]. Distant metastases (DMs) occur in 10–40% of cases, frequently in the lungs (more than 50%), bones (40%), and liver (20%). Metastasis development is related not only to tumour type and stage but also to genetic alterations in tumour cells. These factors are therefore responsible for poor patient outcomes despite radical treatment [18–20].
Currently, the value of genetic analysis with next-generation sequencing (NGS) is particularly highlighted in SGCs. This will not only improve the knowledge about the molecular background of the pathologies but also enable the introduction of targeted therapies, especially for recurrent diseases, advanced stages, and drug-resistant cases [16, 21–24]. Additionally, it might be a pivotal tool in differential diagnosis, especially in ambiguous cases [25]. A summary of the clinical characteristics of SGCs with respect to incidence, histological subtype, predominant location, and survival is presented in Table 1. The most common genetic rearrangements in SGCs are listed in Table 2. The purpose of this paper was to review genetic variations, including novel findings, in the most known histopathological types of SGCs.
Table 1
Clinicopathological features of salivary gland malignancies
| Parameters | Mucoepi-dermoid carcinoma | Adenoid cystic carcinoma | Acinic cell carcinoma | Salivary duct carcinoma | Myoepi-thelial carcinoma | Epithelial-myoepithelial carcinoma | Secretory carcinoma | Carcinoma ex-pleomorphic adenoma | Clear cell carcinoma | Intraductal carcinoma | Adeno-carcinoma | Poly-morphous adeno-carcinoma | Micro-secretory adeno-carcinoma |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Histopathological variant/growth pattern | Oncocytic Clear-cell Sclerosing Low-grade Intermediate-grade High-grade | Cribriform Tubular Solid | Solid papillary-cystic Follicular Microcystic Low-grade Intermediate-grade High-grade | Cribriform Solid Cystic Papillary | Solid Trabecular Reticular | Sebaceous Oncocytic Double-clear | Microcystic Tubular Solid | Myoepithelial carcinoma Salivary duct carcinoma Adenoid Epithelial-myoepithelial carcinoma | Single cells Nested Solid Sheet-like Cords Trabeculae | Intercalated duct type Apocrine Hybrid Oncocytic Low-grade Intermediate-grade High-grade | Variety of growth patterns | Lobular Trabecular Microcystic Cribriform Papillary | Variety of growth patterns |
| Incidence per 1,000,000 (% of all SGC) | 0.62–1.80 (9–30) | 0.41–1.72 (6–25) | 0.41–1.73 (6–17) | 0.27–0.69 (4–10) | 0.14–0.83 (2–12) | 0.34 (< 5) | 0.14–0.27 (2–4) | 0.20–1.10 (3–16) | < 300 cases were described | < 200 cases were described | 0.14–1.24 (2–18) | ~1% | A few cases |
| Predominant location Other location | Major salivary glands (90% in the parotid glands) | Submandibular glands or minor salivary glands Lung, breast | Major salivary glands (87% in the parotid glands) | Major salivary glands | Major salivary glands | Major salivary glands | Major salivary glands | Major salivary glands | Intraoral minor salivary glands | Major salivary glands | Major salivary glands | Minor salivary glands | Minor salivary glands |
| 5-year survival (%) | 37.5–100 | 60–90 | 33–96 | 20–50 | 50–64 | 80–96 | ~95 | 25–96 | No data | No data | 43–81 | 75–100 | No data |
| References | [2, 26–28] | [2, 51, 185, 186] | [2, 73, 185, 186] | [2, 185] | [88, 185, 187] | [143, 185] | [152, 154, 183, 185] | [2, 69, 185, 186] | [168] | [176, 185] | [69, 185] | [2, 105, 185, 188] | [189] |
Table 2
The most frequent genetic alterations in salivary gland carcinomas
| Histopathological type | Fusions | Other genetic changes | References | |
|---|---|---|---|---|
| Mucoepidermoid carcinoma (MEC) | CRTC1-MAML2, 56–88% | TP53, 21–42% CDKN2A, 42–56% CDKN2B, 31% BAP1, < 21% PIK3CA, 17–21% HRAS, < 14% | Saade et al. [31] Kang et al. [34] Seethala et al. [35] Zerdan et al. [47] Wang et al. [48] Morita et al. [49] | |
| Acinic cell carcinoma (AcCC) | SCPP gene cluster – NR4A3, > 80% | CDKN2A/B high percentage in high-grade tumours and metastases cases ATM, 7–14% PTEN, 10–12% | Haller et al. [75] Dogan et al.[78] Ross et al. [69] | |
| Adenoid cystic carcinoma (AdCC) | MYB-NFIB, 60–80% MYBL1-NFIB, MYBL1-YTHDF3 | NOTCH signalling pathway, ~ 40% (NOTCH1, 26%) R/M primary tumours, ~ 13 (NOTCH1, 8.5) KDM6A, ~ 15 BCOR, 13–17 ARID1A, 7–14 | Wagner et al. [61] Ho et al. [59] Lee et al. [66] Ross et al. [69] Wang et al. [68] | |
| Adenocarcinoma (AC) | ||||
| Polymorphous adenocarcinoma (PAC) | PRKD1 hotspot mutation, 50–73% | Andreasen et al. [108] Weinreb et al. [107] | ||
| Cribriform adenocarcinoma (CA) | PRKD1-3 fusions, > 80% | Weinreb et al. [115] | ||
| Microsecretory adenocarcinoma (MiAC) | MEF2C-SS18, ~ 90% | Skálová et al. [39] | ||
| Basal cell adenocarcinoma (BCAC) | CYLD mutation, 29% | Rito et al. [190] | ||
| Mucinous adenocarcinoma (MAC) | AKT1 E17K mutation, 100% TP53 mutation, 88% | Rito et al. [190] Rooper et al. [191] | ||
| Salivary duct carcinoma (SDC) | TP53, 39–60% HRAS, 11–49% ERBB2, 10–100% NF1, 7–20% PIK3CA, 19–47% PTEN, 6–13.5% AR overexpression | Dalin et al. [126] Ku et al. [140] Kohsaka et al. [136] Dogan et al. [127] Mueller et al. [123] | ||
| Myoepithelial carcinoma (MECA) de novo MECA ex PA | TGFBR3-PLAG1, 25% FGFR1-PLAG1, 29% | Various copy number alternations | Dalin et al. [88] | |
| Epithelial-myoepithelial carcinoma (EMC) | HRAS, 27–87% PIK3CA, 22–40% AKT1, 6.5–20% | Urano et al. [146] Grünewald et al. [148] Chiosea et al. [149] Nakaguro et al. [150] | ||
| Secretory carcinoma (SC) | ETV6-NTRK3, > 95% | Baněčková et al. [192] | ||
| Carcinoma ex-pleomorphic adenoma (CA ex PA) | PLAG1/HMGA2 rearrangements | TP53, 55–100% ERBB2, 39–57% PIK3CA, 8–42% HRAS, 4–23% | Stenman et al. [72] Dalin et al. [88] Chiosea et al. [128] Grünewald et al. [141] Dogan et al. [127] Kohsaka et al. [136] | |
| Clear cell carcinoma (CCC) | EWSR1-ATF1, > 90% | Antonescu et al. [170] | ||
| Intraductal carcinoma (IC) | RET rearrangements, ~ 45% NCOA4-RET (mainly in intercalated subtype) MYO18A-ALK | HRAS PIK3CA High percentage (only in apocrine subtype) | Skálová et al. [179] Weinreb et al. [180] Hsieh et al. [182] Majewska et al. [183] | |
A comprehensive literature search was performed in the PubMed database. We analysed the full texts of the articles published in English in the period 1984–2024. The exclusion criteria were as follows: languages other than English, only abstracts available, papers concerning HNC holistically without specific analysis of SGCs, and analysis of malignant transformation of benign lesions, e.g. pleomorphic adenoma.
The search was performed with the following keywords: “salivary gland carcinoma”, “genetic alterations’’, “molecular abnormalities”, “NGS”, “targeted therapy”, “precision therapy”, “mucoepidermoid carcinoma”, “acinic cell carcinoma”, “adenoid cystic carcinoma”, “carcinoma ex-pleomorphic adenoma”, “Ca ex PA”, “adenocarcinoma’’, “salivary duct carcinoma”, “myoepithelial carcinoma”, “epithelial-myoepithelial carcinoma”, “secretory carcinoma”, “polymorphous adenocarcinoma”, “cribriform adenocarcinoma”, “microsecretory adenocarcinoma”, “basal cell adenocarcinoma’’, “mucinous adenocarcinoma”, “clear cell carcinoma”, and “intraductal carcinoma”.
The results of the search are presented in relation to the histopathological types of SGCs.
Mucoepidermoid carcinoma
Mucoepidermoid carcinoma is the predominant salivary gland neoplasm and is detected in more than 30% of all salivary malignancies [26]. Generally, it is characte-rised by gradual growth, rare recurrence, and favourable patient outcomes. However, this type of cancer can be highly heterogeneous and can present as low-, intermediate-, or high-grade cancer, with the latter being associated with poor outcomes. Additionally, the mean age at diagnosis is lower than that of other subtypes and ranges from 45 to 49 years [2, 26–29].
Chromosomal translocation t(11;19)(q14-21; p12-13) is unique for MEC and results in CREB regulator transcriptional coactivator (CRTC1) (also known as MECT1)-mastermind-like transcriptional coactivator 2 (MAML2) oncogene fusion. It has been detected in more than 80% of patients with this cancer subtype. This alteration leads to cell proliferation and survival through autocrine amphiregulin (AREG)/epidermal growth factor receptor (EGFR) signalling [30–35]. Chen et al. revealed that aberrantly activated AREG-EGFR signalling in CRTC1-MAML2-positive MEC cells made them highly sensitive to EGFR inhibition, suggesting benefit from EGFR-targeted therapies, e.g. cetuximab [36]. However, the results of further studies were unsatisfactory, and Ni et al. proposed simultaneous therapy with erlotinib-EGFR inhibitors and Notch inhibitors as more effective [32]. Since MAML2 is involved in NOTCH signalling pathway activation [33, 37, 38], this drug combination becomes more target specific. The other translocation, t(11;19)(q21;q26), results in a CRTC3-MAML2 fusion product that is detected in 6% of cases [30, 39, 40]. Another rare change is the translocation t(6;22)(p21;q12), which promotes ESWR1‒POU5F1 fusion [40]. Previously, the CRTC1-MAML2 fusion product was considered a positive prognostic factor [41–43]. However, further research did not reveal significant differences in survival between patients with and without the translocation [31, 44, 45]. In contrast, Anzick et al. revealed that adverse outcomes in patients with translocations might be related to other genetic alterations, such as CDKN2A deletion [46]. How-ever, copy number variations (CNVs) and somatic mutations associated with this alteration have not been frequently analysed in MEC. Zerdan et al. performed NGS analysis of 118 MEC tumours and reported CDKN2A abnormalities in 53% of the cohort. Other frequent changes included those in TP53 (41%), CDKN2B (31%), BAP1 (19%), PIK3CA (17%), TERT (15%), and HRAS (14,5%) [47]. Similar observations regarding the most common variations were reported by Wang et al. [48]. In contrast, the analysis of comparable sample sizes by Morita et al. revealed that HRAS mutations are rarely detected [49]. On the other hand, Kang et al. reported whole-exome sequencing results for 18 MEC tumours, and the second most frequent variation after TP53 was the POU6F2 gene (17%) [34]. In addition, alterations in BRCA2 and ERBB2 are quite common in MEC (17% and 13%, respectively) [30]. Although NF1 alterations are not frequently detected, Kato et al. reported NF1 and TP53 commutation [47, 50]. However, the significance of these findings remains unclear. Further studies are needed to obtain a more in-depth molecular inquiry into MEC molecular pathogenesis, especially in cases with poor outcomes.
Adenoid cystic carcinoma
Adenoid cystic carcinoma frequently arises in the submandibular or minor salivary glands. Its occurrence in the parotid gland is rare. Although AdCC is known as a histopathological type with indolent growth, it tends to recur, with perineural invasion and DM, especially to the lungs [51–54]. Cases of relapse and metastasis (R/M) are frequently incurable because of a lack of effective systemic therapies, despite ongoing clinical trials. Therefore, there is an urgent need to verify the possibility of using targeted treatment.
The activating neurogenic locus notch homologue protein 1 (NOTCH1) mutation and v-myb avian myeloblastosis viral oncogene homologue (MYB) overexpression are related to AdCC development, progression, perineural invasion, and even chemoresistance, which predisposes patients to unfavourable outcomes [30, 55–58]. In contrast, Ho et al. did not find a correlation between mutational MYBs and either R/M or survival [59]. In approximately 80% of cases, MYB alternations present as the t(6;9)(q22-23;p23-24) translocation, which involves the MYB proto-oncogene and the nuclear factor 1B gene (NFIB) transcription factor, leading to overexpression of the fusion product and worsening the prognosis [30, 60, 61]. MYB NFIB translocation is associated with high MYB expression. This translocation disrupts the MYB 3΄ UTR, a microRNA regulatory site responsible for downregulating MYB. The existence of additional mechanisms for MYB overexpression in AdCC was investigated, revealing alternate rearrangements that translocate super-enhancers in the NFIB and TGFBR3 loci to the MYB locus. The MYB protein binds these super- enhancers, which in turn physically interact with the MYB promoter, drive its overexpression, and establish a positive feedback loop [62].
To emphasise the importance of MYB gene activity, it coordinates the upregulation of pivotal targetable genes involved in several functions related to carcinogenesis, such as apoptosis (API5, BCL2, BIRC3, HSPA8, and SET), cell cycle control (CCNB1, CDC2, and MAD1L1), cell growth and angiogenesis (MYC, KIT, VEGFA, FGF2, and CD53), and cell adhesion (CD34) [63, 64]. Notably, in 35% of MYB-NFIB fusion-negative tumours, MYBL1 alterations were identified [65]. Interestingly, MYB/MYBL1 rearrangements were not very common in R/M AdCCs (22%). In contrast, NOTCH signalling pathway alterations were noted in approximately 40% of R/M cases (with NOTCH1 mutations observed in 26% of these), while only 13% of primary tumours demonstrate increased signalling in the pathway (NOTCH1 mutations in 8.5%) [59, 66].
Notably, Ho et al. also reported frequent alterations in R/M AdCC among genes involved in chromatin remodelling: KDM6A, KMT2C/MLL3, ARID1A, ARID1B, BCOR, MLL2/KMT2D, and CREBBP, with increased frequency compared with primary tumours. TERT promoter mutations were found in > 10% of the R/M patients. Interestingly, NOTCH1 and MYB/MYBL1 fusions are practically undetectable in these lesions [59]. In parallel, Stephens et al., in addition to significant MYB activation, reported SPEN gene alterations (negative NOTCH signalling regulators) in more than 20% of the study cohort [67]. Similar findings regarding NOTCH1, KDM6A, ARID1A, BCOR, CREEB, and TERT have been previously reported. Less frequently detected alterations were in MLL2, RUNX1, PTEN, BAP1, PIK3CA, CDKN2A, ACTB, MGA, CTNNB1, FOXD1, IGFR1, MUC5B, OBSCN, PIK3R1, SPHKAP, TTN FGFR2, and BRAF [68, 69]. In contrast, TP53 mutations are rarely found in AdCCs, including R/M cases. Compared with tumours with favourable outcomes, recurrent and metastatic tumours harbour notably greater loads of mutations. Thus, the options of targeted therapies are quite extensive for verifying their efficiency in advanced stages [56, 70, 71].
Acinic cell carcinoma
The characteristics of AcCC are generally similar to those of MEC. However, some cases of aggressive metastatic AcCC have been reported recently [72–74]. Current knowledge regarding the molecular alterations in AcCC has not yet been properly established.
Haller et al. detected rearrangement t(4;9)(q13;q31), which results in secretory Ca-binding phosphoprotein (SCPP) gene cluster (STATH, HTN1, HTN3, ODAM, FDCSP, and MUC7) and nuclear receptor subfamily 4 group A member 3 (NR4A3) fusion in most tumours of the ana-lysed cohort (more than 80%). The former translocation is unique to AcCC and allows for differentiation of AcCC from mammary analogue secretory carcinoma (MASC), parti-cularly in cases with high-grade transformation. Moreover, the resulting fusion gene acts as an oncogenic driver, with the NR4A3 transcription factor being upregulated due to the translocation of active enhancers from the SCPP gene cluster (which is highly expressed in salivary glands) to the region upstream of NR4A3 [75, 76]. The second most common fusion involves the histatin 3 and Myb/SANT-like DNA-binding domain containing 3 genes (HTN3-MSANTD3) (t(4;9)(q13.3;q31.1)), which have been described in a few cases (4–8%) [75–77]. According to the authors, the former translocation is exceptional for AcCC and provides an effective differential diagnosis of MASC, especially in cases with high-grade transformation. Moreover, NR4A3 might be considered an oncogenic driver through enhancer hijacking, whereby NR4A3 is upregulated [75, 77]. In a recent study, Ross et al. reported CDKN2A and CDKN2B alterations in 76% and 45% of patients with relapses or metastases, respectively [69]. Simultaneously, Dogan et al. performed a genetic analysis and reported that the CDKN2A/B gene changed solely in high-grade tumours (58% of this group), whereas in the disease course with distant metastasis, these rearrangements were found in nearly 90% of the patients [78], confirming them as a negative prognostic factor. Notably, for tumours with identified negative markers, there are targetable treatment options based on CDK4/6 inhibitors, immunotherapy, or DNA methyltransferase inhibitors [79, 80]. Moreover, in advanced AcCC, other genetic changes have also been observed [78]. The most common rearrangements were related to ATM (7–14%), PTEN (10–12%), FBXW7, and TP53 rearrangements, whereas alterations in BRAF, NF1, HRAS, NOTCH1, TERT, ARID2, BIRC3, MTAP, and FAT1 were less common [69, 78]. Importantly, some of these alterations may provide opportunities for utilising precision therapy.
Carcinoma ex-pleomorphic adenoma
Carcinoma ex PA is a rare primary SGC arising from a preexisting PA. It is estimated that 5–15% of benign pleomorphic adenomas undergo malignant transformation to carcinoma (Ca ex PA) [81, 82]. Thus, the detection of the benign part of the tumour might lead to a final misdiagnosis, but rapid growth and other symptoms should indicate suspicion of malignancy [83]. Although salivary duct carcinoma, myoepithelial carcinoma (MECA), and adenocarcinoma not otherwise specified (NOS) are considered the most commonly detected malignant components of Ca ex PA, other types of SGC histopathology have also been described [84–89]. The pleomorphic adenoma gene 1 (PLAG1) and the high-mobility group AT-hook 2 (HMGA2) genes are most frequently altered in both PAs and Ca ex PAs [90], but not typical for primary salivary duct carcinoma (SDC), MECA, or AC. Katabi et al. presumed that rearrangements in these genes were specific to both PA or Ca ex PA and could distinguish Ca ex PA from its de novo counterparts [91]. Nonetheless, further investigations have shown their occurrence in de novo lesions [88]. Carcinoma ex PA tumours have abundant copy number alterations (CNAs) that are suspected to be involved in the malignant transformation from benign lesions. The most common loss of heterozygosity is the amplification of 12q genes (HMGA2, MDM2), deletions of 5q, gains of 8q12.1 (PLAG1) and 8q22.1-q24.1 (MYC), and amplification of 17 chromosomes (ERBB2) [88, 92–94]. Table 3 lists the most commonly detected genetic alterations, including fusions and histopathological subtypes of Ca ex PA, reported in the literature.
Table 3
Malignant component of carcinoma ex pleomorphic adenoma as reported in respective studies
| Gene | Identified malignant component in Ca ex PA | References |
|---|---|---|
| Genes fusions | ||
| CTNNB1-PLAG1 | MECA, SDC ~ 30% | Asahina et al. [193], Skálová et al. [194], Dalin et al. [126] |
| FBXO32-PLAG1 | ND | Bubola et al. [195] |
| FGFR1-PLAG1 | MECA, SDC, ND | Dalin et al. [88], Chiosea et al. [128], Skálová et al. [194], Bubola et al. [195] |
| LIFR-PLAG1 | MECA, SDC | Skálová et al. [194], Dalin et al. [126] |
| MEG3-PLAG1 | ND | Bubola et al. [195] |
| ND4-PLAG1 | MECA | Dalin et al. [88] |
| PLAG1-NFIB | ND | Bubola et al. [195] |
| TGFBR3-PLAG1 | MECA | Dalin et al. [88], Rupp et al. [196] |
| HMGA2-CNOT2 | ND | Bubola et al. [195] |
| HMGA2-NFIB | ND | Bubola et al. [195] |
| HMGA2 fusions | MECA | Dalin et al. [88] |
| Oher PLAG1 fusions | MECA | Dalin et al. [88] |
| HMGA2-WIF1 | ND, Adenoid cystic carcinoma with sarcomatoid transformation, MECA | Persson et al. [92] Katabi et al. [197] |
| ETV6-RET | SC | Smith et al. [198] |
| ZCCHC7-NTRK2 | ND (recurrence and metastatic case) | Pircher et al. [199] |
| Somatic gene mutations | ||
| TP53 | SDC, MECA | Chiosea et al. [128], Grünewald et al. [141], Dogan et al. [127], Rupp et al. [196], Dalin et al. [126], Kohsaka et al. [136], Mueller et al. [123] |
| PIK3CA | SDC, MECA, EMC | Chiosea et al. [128], Dogan et al. [127], Dalin et al. [88], Hallani et al. [144], Dalin et al. [126], Kohsaka et al. [136], Mueller et al. [123] |
| HRAS | SDC, MECA, EMC | Chiosea et al. [128], Dogan et al. [127], Dalin et al. [88], Hallani et al. [144], Dalin et al. [126] |
| ERBB2 | SDC (gain/amp) | Chiosea et al. [128], Dogan et al. [127], Dalin et al. [126], Kohsaka et al. [136], Mueller et al. [123] |
| AKT1 | SDC | Dalin et al. [126] |
| ALK | SDC | Mueller et al. [123] |
| APC | SDC | Dogan et al. [127], Mueller et al. [123] |
| AR | SDC | Dogan et al. [127] |
| ARID1A | SDC | Kohsaka et al. [136] |
| ASXL1 | SDC | Dogan et al. [127] |
| ATM | SDC, MECA | Chiosea et al. [128], Dalin et al. [88], Mueller et al. [123] |
| ATR | MECA | Dalin et al. [88] |
| AURKA | SDC | Dogan et al. [127] |
| BAP1 | SDC | Dogan et al. [127] |
| BRAF | SDC | Chiosea et al. [128], Kohsaka et al. [136] |
| BRCA1 | MECA | Dalin et al [88] |
| BRCA2 | SDC | Dogan et al. [127], Kohsaka et al. [136] |
| BTK | SDC | Dogan et al. [127] |
| CCNE1 | SDC | Dogan et al. [127], Mueller et al. [123] |
| CCND3 | SDC | Mueller et al. [123] |
| CDH1 | SDC | Dogan et al. [127] |
| CDK4 | SDC | Grünewald et al. [141], Mueller et al. [123] |
| CDK6 | SDC | Mueller et al. [123] |
| CDK12 | SDC | Dogan et al. [127] |
| CDKN1B | SDC | Dogan et al. [127] |
| CDKN2A | SDC | Chiosea et al. [128], Mueller et al. [123] |
| CHEK2 | SDC | Mueller et al. [123] |
| CREBBP | MECA, SDC | Dalin et al. [88], Mueller et al. [123] |
| CTCF | SDC | Dogan et al. [127] |
| DNMT1, DNMT3A, NMT3B | SDC | Dogan et al. [127] |
| DOCK7 | SDC | Dalin et al. [126] |
| EGFR | SDC | Dogan et al. [127] |
| EP300 | SDC | Mueller et al. [123] |
| ERBB3 | SDC | Dogan et al. [127] |
| EWSR1 | MECA (clear cell) | Skálová et al. [194] |
| FANCA, FANCC | SDC | Dogan et al. [127] |
| FASN | SDC | Dalin et al. [126] |
| FAT1 | SDC, MECA | Dogan et al. [127], Dalin et al. [88] |
| FAT4 | MECA | Dalin et al [88] |
| FBXW7 | SDC | Dogan et al. [127], Mueller et al. [123] |
| FGFR1 | MECA, SDC | Dalin et al. [88], Dalin et al. [126], Mueller et al. [123] |
| FGFR2 | MECA | Dalin et al. [88] |
| FGFR3 | SDC | Chiosea et al. [128] |
| FGFR4 | SDC | Mueller et al. [123] |
| FH | SDC | Dogan et al. [127] |
| FLCN | SDC | Dogan et al. [127] |
| FOXA1 | SDC | Dalin et al. [126], Kohsaka et al. [136] |
| GATA2 | SDC | Dogan et al. [127] |
| HMGA2 | ND | Persson et al. [92] |
| HNF1A | SDC | Dogan et al. [127] |
| JUN | SDC | Dogan et al. [127] |
| KDR | SDC | Dalin et al. [126] |
| KIT | SDC | Mueller et al. [123] |
| KMT2A | SDC | Dogan et al. [127], Kohsaka et al. [136] |
| KMT2B | SDC | Dalin et al. [126] |
| KMT2C | SDC | Dogan et al. [127], Dalin et al. [126], Kohsaka et al. [136] |
| KMT2D | SDC | Kohsaka et al. [136] |
| LIFR | MECA | Dalin et al. [88] |
| MAP2K2 | SDC | Kohsaka et al. [136] |
| MAP3K1 | SDC | Dogan et al. [127] |
| MDM2 | ND, SDC | Persson et al. [92], Mueller et al. [123] |
| MET | MECA | Dalin et al. [88] |
| MLH3 | SDC | Dalin et al. [126] |
| MML2 | MECA | Dalin et al. [88] |
| MN1 | MECA | Dalin et al. [88] |
| MSH5 | SDC | Dalin et al. [126] |
| MTOR | SDC | Dalin et al. [126] |
| MYC | SDC | Dogan et al. [127] |
| NCOA1, NCOA2 | MECA | Dalin et al. [88] |
| NCOR1 | SDC | Dogan et al. [127], Dalin et al. [126] |
| NF1 | SDC | Dogan et al. [127], Dalin et al. [126], Kohsaka et al. [136], Mueller et al. [123] |
| NOTCH1 | MECA, SDC | Dalin et al. [88], Mueller et al. [123] |
| NOTCH2-3 | SDC | Mueller et al. [123] |
| NSD1 | SDC | Dalin et al. [126] |
| PIK3R1 | SDC | Dogan et al. [127] |
| PTEN | SDC | Chiosea et al. [128], Dogan et al. [127], Kohsaka et al. [136] |
| PTPN11 | SDC | Dogan et al. [127] |
| PTPRS | SDC | Dogan et al. [127] |
| RAD51C | SDC | Dogan et al. [127] |
| RET | SDC | Dalin et al. [126] |
| RICTOR | SDC | Mueller et al. [123] |
| ROS1 | SDC | Mueller et al. [123] |
| RTEL1 | SDC | Dogan et al. [127] |
| SF3B1 | SDC | Dalin et al. [126] |
| SMAD4 | SDC | Dalin et al. [126] |
| SMARCA4 | MECA, SDC | Dalin et al. [88], Dalin et al. [126] |
| TSC2 | SDC | Mueller et al. [123] |
| ZFHX3 | SDC | Kohsaka et al. [136] |
Myoepithelial carcinoma
The incidence of MECA is estimated to be very low, at 2% among all SGCs. Nonetheless, because of the difficulty of proper diagnosis, the actual number of cases is predicted to be greater [10, 95]. The tumour might occur as a de novo lesion or arise from the malignant transformation of a PA or myoepithelioma [96]. These data suggest that MECA ex PAs are more frequently detected than de novo lesions [88, 97]. However, the conclusion regarding which component is characterised by more aggressive behaviour or poorer patient outcomes remains debatable [95, 97–100]. In most cases, this subtype of cancer is associated with adverse patient results, including early local and DM [10, 88, 95]. Myoepithelial carcinoma is one of the most commonly confirmed components of Ca ex PAs [89, 101].
Salivary gland MECA rarely occurs; therefore, few genetic studies of this type are available. Dalin et al. analysed 40 tumours with divisions on either the MECA de novo or the MECA ex PA, as well as cases with and without recurrence. In MECA ex PA, more genetic alterations, including fusions, somatic mutations, and CNVs, were found. According to the authors, CNVs are responsible for the malignant transformation of the PA into the MECA ex PA and are also associated with a worse prognosis. FGFR1-PLAG1 fusion was the most commonly (18%) identified in the MECA ex PA, followed by TGFBR3-PLAG1 but with no evidence of their prognostic value. Furthermore, EWSR1-ATF1 was described only in the MECA de novo, with or without recurrence [88]. In contrast to the research conducted by Skálová et al., EWSR1 rearrangements were found frequently in the clear cell component of MECA both in de novo cases and those arising from the PA, but the fusion partner genes were not identified [102]. In the aforementioned study, PIK3CA was present only in patients without relapse, whereas FGFR2 mutations were found in patients with recurrence [88]. The findings are summarised in Table 4. FGFR2 mutations were also described in 2 patients after radical PA excision, in which the MECA rapidly developed. In both PAs and MECAs (without the PA component), FGFR2 point mutations were confirmed, which might be indicative of an aggressive disease course [103]. Recently, Gandhi et al. reported a novel CTCF-NCOA2 fusion in a single MECA patient [104]. Furthermore, Cormier et al. presented a novel TERT promoter mutation in metastatic MECA ex PA (the tumour was previously misdiagnosed as PA) [9].
Table 4
Genetic rearrangements in the mucoepidermoid carcinoma de novo and the mucoepidermoid carcinoma ex pleomorphic adenoma presented in the study by Dalin et al. in relation to oncological outcomes
Adenocarcinoma
Polymorphous adenocarcinoma
Polymorphous adenocarcinoma (PAC) is a rare, slow- growing malignant tumour. It mainly arises from the minor salivary glands (second most common histopathological type), particularly those localised on the hard palate. There is a higher prevalence in women than in men, and patient outcomes are defined as one of the most favourable outcomes among SGCs [105, 106].
Weinreb et al. revealed a PRKD1 p.E710 hotspot mutation in nearly 73% of tumours, and these observations were not identified in other SGCs. Thus, this alteration is unique to PAC and may be useful for differentiating it from its mimics [107, 108]. Notably, in cribriform adenocarcinoma (CA), PRKD1-3 fusions are the most common. CA is classified as an aggressive variant of PAC with a high predisposition to metastasis [109–112]. Among the fusion partners ARID1A, ATL2, DDX3X, PPP2R2A, PRKAR2A, SNX9, and STRN3 (cases with high-grade transformation) should be mentioned [113–116]. However, the type of genomic alteration is not specific for any AC subtype, and occasionally, either PRKD1-3 fusions or PRKD1 rearrangements are found in PAC and CA, respectively [109]. Therefore, differentiation between these 2 variants with various behaviours might be challenging.
Adenocarcinoma not otherwise specified
Tumours with a histopathological diagnosis of adenocarcinoma NOS constitute a heterogeneous group that has not yet been well characterised. For example, NTRK2-ZCCHC7 and SS18-ZBTB7A fusions have been described [116, 117]. In R/M cases, TP53 (55%), PIK3CA, HRAS, CDKN2A, ERBB2, PTEN, NF1, and ARID1A alterations were observed with considerable frequency [69].
On the basis of genetic pattern analysis, microsecretory adenocarcinoma has been distinguished from NOS. Microsecretory adenocarcinoma harbours MEF2C-SS18 fusion in approximately 90% of cases [39, 118].
The most common alterations in basal cell adenocarcinoma and mucinous adenocarcinoma are shown in Table 1.
Salivary duct carcinoma
Salivary duct carcinoma is one of the most aggressive SGCs, with either early relapse or frequent DM. It is also associated with significant mortality. Predilection in elderly males with a smoking history is usually combined with advanced-stage presentation and parotid gland localisation [119–123]. The estimated morbidity is 5.5–12% [124, 125]. Moreover, SDCs ex PAs have also been detected [122, 126–128]. Table 2 provides genetic information for this subtype.
In addition to the microscopic structure resembling high-grade ductal carcinoma of the breast, SDC is also characterised by the overexpression of human epidermal growth factor receptor 2 (HER2). Instead of oestrogen and progesterone receptor positivity, androgen receptor (AR) expression is detected in 75–98% of cases [122, 126, 129, 130]. Notably, AR is seldom detectable in other SGCs [131]. However, studies are inconclusive regarding the prognostic value of the AR [129, 131, 132]. Nevertheless, Kawakita et al. showed in a retrospective study that the utilisation of HER2-targeted therapy and androgen deprivation therapy significantly improved patients results compared with conventional therapy management [133]. The anti-HER2 therapies that induce improvement in clinical responses in SDC patients use trastuzumab in combination with chemotherapy (i.e. taxanes, capecitabine, carboplatin, eribulin) or with another anti-HER2 targeted agent (i.e. pertuzumab). Further expectations and therapeutic advances are related to novel anti-HER2 drugs such as antibody-drug conjugates (i.e. trastuzumab emtansine, trastuzumab deruxtecan) introduced in this setting [134].
In recent years, genetic knowledge about SDC has increased profoundly, but it still has not been comprehensively investigated. The tumour mutation burden is extremely high in most SDC cases, in contrast to other SGCs. Vos et al. evaluated therapy with nivolumab (anti-PD-1) and ipilimumab (anti-CTLA-4) in patients with metastatic SGC. Although the efficacy was limited in AdCC, with infrequent responses, they found it promising for non-AdCC SGCs, particularly salivary duct carcinomas [135]. Genetic fusions are not recurrent events in this subtype, whereas somatic mutations as well as CNVs are considerably more common [123, 126, 136]. Moreover, most of them provide opportunities for the utilisation of targeted treatment for this unpredictable cancer [30, 127, 137–139]. TP53, HRAS, PIK3CA, and ERBB2 (HER) rearrangements are the most common, and some of them are related to poor outcomes [123, 126–128, 136, 140, 141]. Interestingly, although HRAS mutations constitute the majority of de novo lesions, they are rare in SDC ex PAs [123, 126, 127, 136]. Data regarding the molecular landscape of SDCs are presented in Table 5.
Table 5
The genetic pathways most commonly affected in salivary duct carcinoma
| Pathway | Genes | References |
|---|---|---|
| DNA damage | TP53 (39–60%), ATM, BRCA2, CHEK2, MDM2, MDM4, MLH3, MLH5 | [123, 126, 127, 136, 140, 141] |
| MAPK | HRAS (11–49%), NF1 (7–20%), BRAF, KRAS, NRAS | [123, 125, 126, 127, 128, 136, 137, 140] |
| RTK | ERBB2 (10–100%), ALK, EGFR, ERBB3-4, FGFR1-2, FGFR4, FLT3, JAK2, KDR, KIT, MET, NTRK2, PDGFRA, RET | [123, 126, 127, 136, 137, 140] |
| PI3K/AKT/mTOR | PIK3CA (19–47%), PTEN (6–13.5%), AKT1-3, PIK3R1, RICTOR, RPTOR, TSC2 | [123, 125, 126, 127, 128, 136, 137, 140] |
| Androgen signalling | AR, FASN, FOXA1 | [126, 136] |
| Histone modification | KDM6A, KMT2A, KMT2C, KMT2D, KMT2E, NSD1 | [126, 127, 136, 140] |
| Cell cycle | CDK4, CDK6, CDK12, CDKN1A, CDKN1B, CDKN2A, CCNE1, CCND1-3, RB1 | [123, 126, 127, 136, 140, 141] |
| NOTCH | CREBBP, EP300, FBXW7, NOTCH1-3 | [123, 140] |
| SWI/SNF complex | ARID1A, SMARCA4, SMARCB1 | [123, 126, 127, 136] |
| WNT-β-catenin | APC, CDH1, CTNNB1, FAT1 | [123, 126, 140] |
| Other | ABL1, AURKA, BCOR, CCND1, CCNE1, FLCN, GNAS, HMGA2, IDH1-2, IGFR1, IKBKE, KLF5, AMP, MAP2K1, MAP2K4, MITF, MPL, MYC, PRDM1, SMAD4, SMO, STK11, TNIK, VHL, ZFHX3 | [123, 126, 127, 136, 140, 141] |
| Fusions | ETV6-NTRK3, ABL1-PPP2R2C, BCL6-TRADD, HNRNPH3-ALK, EML4-ALK, RAPGEF6-ACSL6 | [126, 127, 195] |
Epithelial-myoepithelial carcinoma
Epithelial-myoepithelial carcinoma (EMC) is rarely detected, and it was first reported by Donath et al. in 1972. Previously, it appeared under other terminology of adenomyoepithelioma or clear cell adenoma. The tumour consists of a dual cell population that forms a double layer: inner ductal cells and outer myoepithelial cells [142–144]. Notably, various histological subtypes of EMCs exist, including sebaceous, oncocytic, and double-clear subtypes. Thus, the differential diagnosis could pose difficulties [145, 146]. Morbidity predominates in females more than males. Most commonly, the parotid gland is affected, and the tumour is characterised by a high overall survival rate. Although DM rarely occur, relapses are common [143, 147].
HRAS (27–87%) was described as the most frequently mutated gene in EMC [146, 148–150]. In the studies conducted by Urano et al. and Nakaguro et al., these findings were not detected in EMCs ex PAs [146, 150]. In parallel, Hallani et al. did not prove HRAS alterations for de novo EMC [144]. PIK3CA and AKT1 have been reported quite commonly in EMC (22–40% and 6.5–20%, respectively) [146, 148]. CTNNB1, FBXW7, and TP53 rearrangements and SMARCB1 deletions have been reported in single cases (the last 3 in high-grade tumours) [144, 148]. Mäkelä et al. described rare metastatic EMC in a 36-year-old woman, where in addition to HRAS mutation, ARID1B, ATR, CDK12, ERBB4, MAPK1, NANOG, NOTCH2, PIK3R1, and RPTOR alterations were detected [151].
Secretory carcinoma
Secretory carcinoma (SC) (previously known as mammary analogue secretory carcinoma) is a novel salivary gland tumour that was described by Skálová et al. in 2010 [152]. Most of these tumours were previously classified as AcCC [153]. The age at diagnosis is relatively low (mean 45 years), including paediatric patients. There is a greater predilection in men, and the disease course is indolent, with favourable patient outcomes [154, 155].
Secretory carcinoma has a significant histological and molecular resemblance to breast secretory carcinoma. It is characterised by harbouring the same translocation t(12;15)(p13;q25), resulting in the ETV6-NTRK3 fusion gene encoding a chimeric oncoprotein-tyrosine kinase (unlike AcCC) [152, 155, 156]. Other ETV6 fusion partners have also been discovered, including ETV6-MAML3 [157], ETV6-MET [158], and ETV6-RET [157, 159]. Notably, some of these genes remain unknown (ETV6-X) [160]. Recently, other novel fusions, such as VIM-RET [161], CTNNA1-ALK [162], and dual fusion, ETV6-RET and EGFR-SEPT14, were identified in an 18-year-old male [159]. ETV6-NTRK3 and MYB- SMR3B fusions were found in recurrent high-grade submandibular tumours [161]. Only a few studies have analysed genetic rearrangements other than fusions. Na et al. identified pathogenic PRSS1 mutations, mainly in patients with an aggressive disease course and recurrence, whereas other findings were classified as likely pathogenic or of uncertain significance [163]. In contrast, Skálová et al. analysed 3 tumours with high-grade transformation and did not detect the most commonly occurring genetic alterations associated with poor outcomes (TP53, CTNNB1, EGFR, CCND1) [164].
Testing for ETV6-NTRK3 gene rearrangements is critical for SC patients care since entrectinib, an inhibitor of tropomyosin receptor kinase (TRKs), has been reported to be effective and safe in treating solid tumours with NTRK fusion genes. In an integrated analysis of phase 1–2 trials (STARTRK-1, STARTRK-2, and ALKA-372-001) of solid tumours with the NTRK fusion gene, the response rate to the TRK inhibitor entrectinib was 57%, and the median progression-free survival was 11.2 months [165]. Another TRK inhibitor, larotrectinib, is also effective in the treatment of solid tumours with the NTRK fusion gene [166]. Other potential therapies for SC patients with identified oncogenic RET fusions, namely ETV6-RET , are selpercatinib and pralsetinib selective RET inhibitors, currently under preclinical and clinical testing [167].
Clear cell carcinoma
Clear cell carcinoma (CCC) (previously known as hyali-nising clear cell carcinoma) is an indolent low-grade tumour that typically arises from the intraoral minor salivary glands. There is a higher prevalence in females, whereas relapses and metastases are rare [168].
Considering the occurrence of clear cells in other SGCs, differential diagnosis may be a challenge [169]. Antonescu et al. first described genetic rearrangement in the CCC-EWSR1-ATF1 fusion t(12;22)(q13;q12). It occurs in more than 90% of cases, and, being unique for CCC, it is therefore a helpful differentiation tool [170]. EWSR1-CREB1, EWSR1-CREM, and SMARCA2-CREM fusions have been reported in single cases thus far [171–173].
Intraductal carcinoma
Intraductal carcinoma (IC) is a rare salivary gland tumour that affects mainly the parotid gland, with features similar to mammary atypical ductal hyperplasia or ductal carcinoma in situ of the breast [174, 175]. Recent studies have classified 4 distinctive subtypes: intercalated duct type, apocrine, hybrid, and oncocytic [176].
RET rearrangements, including recurrent NCOA4-RET (intercalated, oncocytic, seldom hybrid), TRIM27-RET (hybrid, apocrine), and TRIM33-RET (oncocytic) rearrangements, have been detected [177–179]. In contrast, RET gene altera-tions have not yet been confirmed in the apocrine subtype [180].
The relationship between IC and SDC remains controversial, even though they are considered diverse entities. Intraductal carcinoma, especially invasive apocrine IC, is a precursor for more aggressive cancers, such as SDC [174, 176, 180]. Nevertheless, this issue requires further investigation. Molecular evidence of resemblance to SDC revealed a high occurrence of HRAS and PIK3CA hotspot mutations in apocrine IC [174, 180–182]. Additionally, ATM, SPEN, and TP53 mutations and either DFFA-ARID1A or KIF13B-EPB41L4B fusions were found in this subtype [174,180]. In parallel, BRAF V600E mutations in the oncocytic subtype and novel fusions of TUT1-ETV5 and KIAA1217-RET in intercalated duct variants and hybrid intercalated duct tumours with invasive growth have also been identified [178, 179].
Furthermore, Majewska et al. reported an MYO18A-ALK fusion in intercalated duct-type IC in elderly patients after radical excision and no disease relapse during follow-up [183].
Recently, Watanabe et al. presented a case of a 59-year-old male with high-grade intercalated-type IC and DM. Despite radical excision and postoperative radiotherapy, the patient developed multiple DM. Genetic analysis revealed a CTNNA1-ALK fusion and TP53 mutation. Despite further ALK-TKI therapy, the patient’s condition declined, and NGS analysis of the blood samples revealed a novel PIK3CA mutation (ALK fusion was not detected). The importance of this shift remains uncertain. Nevertheless, treatment failure might be related to novel alterations and the predominance of other abnormalities in recurrent tumour tissue [184].
Conclusions
Salivary gland carcinomas are rare entities with unpredictable disease courses. The diversity of both the histological architecture and molecular alterations is distinct among individual subtypes, which leads to diagnostic difficulties. Moreover, because of the rare incidence of SGCs, multicentre clinical trials are urgently needed to provide targeted therapeutic options. Currently, the value of gene-tic analysis has been highlighted, particularly in terms of the possibilities of precision therapies and in light of the insufficient effectiveness of standard treatment options. Knowledge of the molecular landscape of SGC, especially related to outcome predictors, will provide novel and precise methods for diagnosis and therapy in the future.
