Fusion proteins in head and neck neoplasms: Clinical implications, genetics, and future directions for targeting

ABSTRACT

Fusion proteins resulting from chromosomal rearrangements are known to drive the pathogenesis of a variety of hematological and solid neoplasms such as chronic myeloid leukemia and non-small-cell lung cancer. Efforts to elucidate the role they play in these malignancies have led to important diagnostic and therapeutic triumphs, including the famous development of the tyrosine kinase inhibitor dasatinib targeting the BCR-ABL fusion. Until recently, there has been a paucity of research investigating fusion proteins harbored by head and neck neoplasms. The discovery and characterization of novel fusion proteins in neoplasms originating from the thyroid, nasopharynx, salivary glands, and midline head and neck structures offer substantial contributions to our understanding of the pathogenesis and biological behavior of these neoplasms, while raising new therapeutic and diagnostic opportunities. Further characterization of these fusion proteins promises to facilitate advances on par with those already achieved with regard to hematologic malignancies in the precise, molecularly guided diagnosis and treatment of head and neck neoplasms. The following is a subsite specific review of the clinical implications of fusion proteins in head and neck neoplasms and the future potential for diagnostic targeting.

Introduction

Fusion proteins are products of gene fusions resulting from chromosomal rearrangements that are capable of acting as potent oncogenes. They operate via a myriad of mechanisms, including kinase activation or the deletion of regulatory microRNAs.^1-2 Their clinical relevance is evident in the insight they provide into early tumorigenesis as well as their role in facilitating the development of novel targeted therapies and diagnostic applications.

Gene fusions were first identified in hematological malignancies with the initial description of BCR-ABL in chronic myeloid leukemia in 1960.³ Discovery of this f usion led to the development of targeted first-line therapies such as the tyrosine kinase inhibitor dasatinib.⁴ Numerous fusion genes have subsequently been described in solid tumors.³

Various events are capable of generating oncogenic fusion proteins, including inter-chromosomal translocations or intra-chromosomal rearrangements such as translocations, deletions, and inversions⁵ Interestingly, viruses are also capable of generating fusions harboring oncogenic activity.⁶

Fusion proteins are being recognized for their relevance in head and neck neoplasms and have thus far been identified in malignancies originating from the nasopharynx, salivary glands, maxilla, auditory meatus, lacrimal glands, thyroid, esophagus, and midline head and neck structures.^1,7,9 They provide logical targets for therapies that may benefit patients with unresectable head and neck malignancies, while providing valuable contributions to our understanding of their pathogenesis, diagnosis, and prognosis.

The following provides a subsite specific review of the clinical implications of fusion proteins in head and neck neoplasms and the future potential for diagnostic targeting.

Salivary gland neoplasms

Salivary gland tumors represent a diverse group of malignancies, and a myriad of subtypes with important clinical and biological implications have been described.¹⁰ They often harbor recurrent chromosomal translocations, thereby generating oncogenes that influence their pathogenesis.⁷ Adenoid cystic carcinoma (ACC), mucoepidermoid carcinoma (MEC), polymorphous low-grade adenocarcinoma (PLGA), mammary analog secretory carcinoma (MASC), hyalinizing Clear Cell Carcinoma (HCCC), and cribiform adenocarcinoma of minor salivary gland origin (CAMSG) are among the subtypes of salivary gland cancers found to harbor specific fusion proteins.¹⁰

Polymorphous low-grade adenocarcinoma (PLGA)

PLGA is a salivary gland tumor with a predilection to the palate that occurs more often in women than in men. It carries an excellent prognosis overall, tends to be small and easily treatable by surgery, and rarely causes death due to disease. It occasionally metastasizes to cervical lymph nodes, but rarely metastasizes hematogenously. CAMSG was most likely previously categorized under PLGA, but has been recently acknowledged as its own entity.¹⁰ It too carries an excellent prognosis, however it differentiates itself from PLGA by its predilection for the base of the tongue, frequent lymph node metastasis at the time of presentation, and its histological characteristics reminiscent of papillary thyroid cancer.¹⁰

Novel recurrent PRKD fusions, including ARID1A-PRKD1 and DDX3X-PRKD1, have recently been identified in the PLGA/CAMSG spectrum. PRKD1 encodes a kinase and, along with PRKD2 and PRKD3, is involved in signal transduction in the diacylglycerol/protein kinase C pathway, while ARID1A is a gene involved in chromatin remodeling whose incapacitation results in DNA repair deficiencies.¹⁰ Fusion between the 2 results in a truncated PRDK1 protein. DDX3X is also involved in DNA repair and has been implicated in medulloblastoma and breast cancer. Although the functions of these genes are known, the mechanisms of the fusions in which they participate remain poorly understood. Weinreb et al found that the majority of tumors harboring these fusions were either CAMSGs or indeterminate cases (tumors within the PLGA spectrum that demonstrated morphology unusual for PLGA), raising the possibility of a distinction between CAMSG and PLGA based on the presence of fusion proteins.¹⁰

Adenoid cystic carcinoma (ACC)

In contrast to PLGA and CAMSG, adenoid cystic carcinoma (ACC) of the salivary glands is a rare malignancy that carries a poor long term prognosis, with a tendency to grow slowly, resist treatment, and recur distantly.^1,11 Various histologic subtypes of ACC exist with different clinical behaviors. Of the cribiform, tubular, and solid patterns, tumors with solid histology will typically be associated with the worst prognosis and distant metastasis.¹²

Over-activation of the MYB oncogene, a transcription factor that plays a major role in regulating cell proliferation, differentiation, and survival, has been described as the hallmark of ACC.^1,11,13 This event often results from a t(6;9)(q22-23;p23-24) chromosomal translation, which generates a fusion between MYB and the transcription factor gene NFIB.¹¹ As a consequence of the MYB-NFIB fusion, MYB loses the portion of its 3′ end that contains micro-RNA binding sites required for the ability of the mIR-150 and miR-15a/16 regulatory micro-RNAs to downregulate its activity.¹ The fusion therefore disrupts MYB inhibition. It is present in the majority of ACCs and is not observed in other salivary tumors.^13,14,15

The MYB-NFIB fusion may have other consequences that contribute to ACC tumorigenesis. The t(6;9) translocation that generates the fusion creates a complex pattern of breakpoints and genomic imbalances affecting 6q and 9p. The rearrangements impacting 6q have been found to result in the loss of 6q24.1-6q27, a region that harbors the 2 candidate tumor suppressor genes LATS1 and PARK2.¹³ Persson et al found 6q24.1-q25.2, 6q26, and 6q27 to be deleted in 75% of high grade ACCs.¹³

A recent case series by Rettig et al explores the clinical implications of tumor fusion status and finds that ACC tumors arising from minor salivary glands are more likely to harbor the MYB-NFIB fusion and that female gender is associated with positive MYB-NFIB tumor status.¹¹ While trends toward greater tumor recurrence, increased nodal metastasis, and decreased 5 year disease free survival were observed in MYB-NFIB positive tumors, these endpoints did not achieve statistical significance. Furthermore, fusion status rendered no difference in overall survival. Therefore, it is still unclear what prognostic significance the MYB-NFIB fusion carries and whether it is the fusion itself or MYB overexpression, which can result from mechanisms other than gene fusion, that influences outcomes.¹¹

To date, there are no clinically available agents available for which to target the MYB-NFIB fusion gene, or the MYB pathway.¹⁶

Mucoepidermoid carcinoma (MEC)

Mucoepidermoid carcinoma (MEC) is the most common salivary gland malignancy, responsible for over 30% of malignant neoplasms of the major/minor salivary glands,^17,18,19 and is thought to arise from precursor cells in excretory and intercalated ducts of the salivary gland.²⁰ Interestingly, MEC can also originate from the thyroid, esophagus, and bronchial glands.⁷ It tends to occur in the fifth decade of life and is slightly more common in women than in men.²⁰ Various grades of MEC exist, and staging systems are based on cell type differentiation, cyst formation, and cytomorphologic changes.^21-24

Many MEC will harbor a unique translocation t(11;19)(q21;p13) which produces a fusion transcript of the mucoepidermoid carcinoma translocated-1 gene (MECT1 [also known as CRTC1]) with the mastermind-like gene (MAML2),²⁵ although MECs positive for this fusion may alternatively demonstrate other translocations such as t(11;17) and t(11;13), trisomies, or even seemingly normal karyotypes.⁷ MECT1 is a member of the CREB (cAMP-response element binding protein) regulated transcription coactivator family that co-activates CREB mediated transcription, whereas MAML2 is a member of the Mastermind-like family of nuclear proteins that co-activate Notch receptors.^7,9 The MECT1-MAML2 fusion encodes a protein in which the CREB-binding domain of MECT1 is joined to the transactivation domain of MAML2, whose oncogenic capability has to do with aberrant activation of the cAMP/CREB pathway and Notch signaling.⁹ Downstream, it upregulates amphiregulin (AREG), which in turn acts in an autocrine fashion to bind to EGFR, thereby initiating signaling instigating MEC cell proliferation and survival. The significance of this mechanism is supported by the observation that AREG depletion results in decreased MEC cell growth and survival in vitro and in vivo.⁷ Furthermore, the anti-EGFR antibody Cetuximab dampens the proliferation of MECT1-MAML2 fusion positive MEC cancer cells by disrupting AREG-EGFR signaling.⁷ These data provide merit to the potential role for EGFR inhibitors in the treatment of patients with advanced and unresectable fusion-positive MEC.⁷

The MECT1-MAML2 fusion has been evaluated for its utility in predicting the biological behavior of MECs. Behboudi et al found that patients with fusion positive tumors tend to be significantly younger than those with fusion negative tumors (median 48 vs. 73 years). Histologically, the majority of fusion positive MECs were found to be highly differentiated tumors, whereas fusion negative tumors tended to be poorly differentiated. Moreover, fusion positive status seemed to render a more favorable prognosis, with a significantly reduced propensity for local recurrence, metastasis, or tumor related death.^20,26,27,28 Notably, the estimated median survival for fusion positive patients was found to be greater than 10 years, compared to 1.6 y for fusion negative patients.²⁶ A recent investigation by Tirado et al also noted a significant correlation between fusion negative tumors and distant metastasis, but found no significant difference between age, gender, tumor site, clinical stage, histological grade, nodal status or perineural invasion based on fusion status.²⁶ Moreover, the MECT1-MAML2 fusion was identified in all MEC grades, thereby challenging the association between fusion transcript positivity and low grade MECs.²⁶ Others have noted the MECT1-MAML2 fusion transcript can be identified in 38-81% of salivary MECs, the majority of which are low and intermediate grade. In addition, this transcript is very specific, with several studies suggesting translocation-negative high-grade MEC to actually be considered a different class of tumor.^27,29 The EWSR1-POU5F1 fusion has been reported in undifferentiated variants of MEC that are MECT1-MAML2 negative.^9,26 Other MEC-related fusions include the CRTC3-MAML2 fusion, which is noted to be present in 6% of cases, though lack an association with disease-free survival.³⁰

Interestingly, MECT1-MAML2 has also been identified in Warthin's tumor, a benign salivary gland neoplasm occurring mostly in the parotid gland.³¹ A recent investigation identified this fusion in Warthin's tumors with synchronous MEC, suggesting that it might actually represent an event associated with the development of a subset of Warthin's tumor that may be prone to malignant transformation.³¹

Mammary analog secretory carcinoma (MASC)

Histologically similar to secretory carcinoma of the breast, MASC is a comparatively new salivary tumor first noted in 2010.³² Both tubular and microcystic patterns have been described, and may occasionally be mistakenly identified as acinic cell carcinoma.^33,34 The immunohistochemical signature of MASC is unique from other salivary tumors given S100, cytokeratin, mammaglobin, and vimentin positivity.³⁵ Behavior is locally infiltrative, considered low-grade, with minimal rates (<20 %) of recurrence and metastasis, though high-grade transformation has been described.³⁵

MASC has a t(12;15)(p13;q25) translocation which results in a fusion product between the ets variant 6 (ETV6) and neurotrophic tyrosine kinase receptor type 3 (NTRK3) genes which is unique in the context of head/neck salivary tumors. ETV6 is a transcriptional regulator, and NTRK3 is a membrane receptor kinase activating cell proliferation and survival.³² The functional significance of this fusion is still being elucidated though is thought to be potentially related to the increased transcription of signal transducer and activator of transcription molecule 5a (STAT5a) noted in MASC.^32,36 The ETV6-NTRK3 fusion product is also able to activate the PI3k-Akt and Ras-MAP pathway.³⁷ Additional investigation regarding targeting of this fusion product is needed, though of interest is the treatment response noted when using tyrosine inhibitors for ETV6-NTRK3 positive lymphomas.³⁸

Pleomorphic adenoma

Several fusion genes have been identified in pleomorphic adenoma (PA), a benign salivary gland neoplasm. The majority of fusions observed in this entity involve the transcription factor PLAG1, which is most commonly partnered with CTNNB1 as a result of t(3;8)(p21:q12) rearrangement. These fusions commonly result in transcriptional upregulation of PLAG1 and protein overexpression as a result of substitution of the PLAG promoter region by the promoter of the partner gene. Three additional fusion genes involving HMGA2, a chromatin remodeling gene, have been identified in PA. Notably, the HMGA2-WIF1 fusion has been identified in carcinoma ex pleomorphic adenoma, thereby implicating a fusion protein in a benign neoplasm's potential for malignant transformation.⁸

Hyalinizing clear cell carcinoma (HCCC)

Hyalinizing Clear Cell Carcinoma (HCCC) is a rare salivary gland tumor which is low grade in nature, with rare metastasis, and overall good prognosis.³⁷ A translocation in Ewing sarcoma RNA-binding protein 1 (EWSR1) is noted in over 80% of HCCC, and may be of use diagnostically given the common histologic findings of clear cells in other salivary tumors. A translocation of t(12;22)(q13;q12) results in a EWSR1-ATF1 fusion transcript.³⁹ The specific function of this fusion product has not yet been described in detail, though interestingly it is present in odontogenic clear cell carcinomas as well.³⁵

Thyroid neoplasms

Kinase fusions play a prominent role in thyroid neoplasms, as evidenced by identification of a high percentage of recurrent fusions involving the ALK, NRTK, ROS, BRAF, RET and FGFR gene families.^2,4 Although the kinase fusions reported in in thyroid malignancies are numerous, they operate via mechanisms with a common underlying theme. In the majority of cases, a fusion partner provides dimerization domains that constitutively activate a kinase, thereby contributing to tumor development.²

Thyroid carcinoma has also been shown to harbor a high percentage (70.3%) of in frame fusions compared to other malignancies.⁴ Yoshihara et al found that tumor samples with recurrent in-frame fusion transcripts harbored less frequent gene mutations compared to samples without recurrent in frame fusion transcripts.⁴ This discrepancy achieved statistical significance with thyroid cancer as well as head and neck squamous cell carcinoma, suggesting that fusions are indeed driving tumorigenesis in these neoplasms.

Recurrent fusions involving the anaplastic lymphoma kinase (ALK) have been detected in thyroid cancers, including STRN-ALK and the more recently identified GTF2IRD1-ALK.^2,4 Interestingly, ALK fusions have also been described in esophageal cancer.⁴ Of note, they can be targeted by ALK inhibitors,⁴ highlighting the utility of fusions in guiding rational therapy for head and neck neoplasms.

Fusions involving RET, a tyrosine kinase found on chromosome 10q, are common in papillary thyroid carcinoma (PTC), found in approximately 35% of PTCs in the US adult population.⁸ RET is also often activated by mutations in medullary thyroid cancer, and RET inhibitors are approved for treatment of this malignancy.⁴ Many RET fusions have been described, most of which follow the activation paradigm whereby the fusion partner provides dimerization domains that constitutively activate the kinase.⁸ Some of the RET fusions recurrently identified in thyroid cancer involve other genes located on chromosome 10 via paracentric chromosomal inversion, including CCDC6and NCOA4.⁸ NCOA4-RET is frequently associated with prior radiation therapy, while CCDC6-RET occurs with greater frequency in PTCs with classic histological features⁸. Novel RET fusion partners, including AKAP13, FKP15, SPECC1L, and TBLXR1 have recently been identified.² Of these, TBLXR1 contains a Lis-homology (LisH) motif in the coding region, which likely activates the RET kinase via dimerization, while the rest of the fusion partners contain dimerization-capable coiled-coil motifs.² The abundance of recurrent RET activating fusions identified in thyroid cancers emphasizes a therapeutic rationale for treatment with RET inhibitors.⁴

NTRK1 and NTRK2 fusions have been consistently identified in PTC.² Moreover, the ETV6-NTRK3 fusion, previously identified in several non-thyroid malignancies such as secretory breast cancer, congenital fibrosarcomas, mesoblastic nephromas, and AML, has recently been observed in thyroid neoplasms.⁵ This fusion may contribute to the oncogenic properties of thyroid cancer by impacting the PI3K/AKT and RAS-MAPK pathways.⁵ NTRK fusions may play a role in other head and neck cancers, as evidenced by the recent identification of the novel PAN3-NTRK2 in head and neck squamous cell carcinoma.² These fusions are also considered to be druggable, inviting the use of pan-NTRK inhibitors.⁴

Interestingly, many fusions involving BRAF have been identified in thyroid neoplasms.^4,5,8 BRAF activating point mutations play a well-known role in PTC, occurring in up to 40%, and have been described in radiation induced thyroid cancer.⁸ Therefore, distinct mechanisms, i.e., fusions and point mutations, can lead to the same consequence of constitutive serine-threonine kinase activation resulting in pathological mitogen activated protein kinase (MAPK) signaling and tumor formation.⁸ Among the BRAF fusions identified in thyroid malignancies is the AKAP9-BRAF fusion found in PTC, which results from a paracentric inversion involving chromosome 7q.⁸ Others include ZC3HAV1-BRAF and FAM114A2-BRAF fusions. These function similarly to RET fusions by encoding 5′ protein partners that contribute zinc finger or coiled-coil motifs, which likely generate constitutively active BRAF dimers capable of driving tumorigenesis.² Both fusions are poorly responsive to RAF inhibitors, however are sensitive to downstream inhibition via the MEK pathway.² MKRN1-BRAF is a novel fusion in which the BRAF coding sequences involve the entire kinase domain and would therefore be expected to demonstrate kinase activity and be potentially transforming.⁵

In addition to BRAF fusions, the AGGF1-RAF1 fusion has been identified in multiple PTCs.² It functions similarly to BRAF fusions via RAF1 activation by the end terminal coiled coil dimerization motif contained on AGGF1, resulting in constitutively active RAF1 dimers resistant to RAF inhibitors.² TFG-MET is another PTC fusion that follows this familiar activation paradigm, and therefore represents another druggable fusion.²

Recurrent non-kinase fusions have also been described in thyroid neoplasms. HEPHL1-PANX1 and METTL1-FAM53B are intra-chromosomal fusions arising from fusion partners in the same strand and orientation and therefore may result from transcriptional read-through, deletion, or translocation. Other recurrent thyroid fusions discovered likely result from translocation and include KIAA126-ARL17B, PPIP5K1-CATSPER2, and RHOBTB2-PEBP4. While the functional significance of these fusion transcripts is unclear, several of the genes participating, including PANX, FAM53B, RHOBTB2, and PEBP4 have known oncogenic or tumor suppressive functions.⁵

Whereas PTC is characterized by the presence of RET and NTRK1 tyrosine kinase fusion genes, roughly 30% of follicular thyroid carcinoma (FTC) and 38-50% of follicular variant papillary thyroid carcinoma (FVPTC) harbor the PAX8-PPARG1 fusion, a product of the t(2;3)(q13;p25) chromosomal translocation.⁸ PAX8 is a transcription factor that contributes to the development of thyroid follicular cells, whereas the peroxisome proliferator-activated receptor gamma nuclear receptor (PPARG1) is involved in lipid and carbohydrate metabolism, energy balance, eicosanoid signaling, and adipocyte differentiation. The functional significance of the PAX8-PPARG1 fusion has yet to be fully elucidated, but it may act as a negative regulator of PPARG1.⁸ Interestingly, the PAX8-PPARG1 fusion is found in a subset of follicular adenomas, giving rise to the question whether these benign tumors may evolve into carcinomas.⁸ The presence of fusions in these benign entities highlights the pleotropic role of fusion genes in the development of both malignant and benign tumors.

A PPARG1 fusion with an alternate partner cyclic AMP response element binding protein 3 (CREB3) has recently been identified in 3% of follicular thyroid carcinomas.⁸ This discovery resulted from the observation that several follicular thyroid carcinomas demonstrated overexpression of PPARG1 while testing negative for the PAX8-PPARG1 fusion.⁸ While the pathologic role of CREB3L2-PPARG1 remains to be fully characterized, the fusion has been found to induce proliferation and interfere with CRE related transcription in normal thyroid cells.⁴⁰

It is notable that the MECT1-MAML2 fusion found in salivary gland MECs and Warthin's tumors is also found in thyroid MEC, a rare entity thought to originate from follicular epithelial cells.²⁶ Consistent with the data suggesting that fusion negative salivary gland MECs may represent a subset of biologically aggressive tumors, the 2 fusion negative thyroid MECs examined by Tirado et al demonstrated distant metastasis, while the fusion positive one did not.²⁶

Nasopharynx

Nasopharyngeal carcinoma (NPC) is a common head and neck malignancy that demonstrates geographical variation in incidence. Although Epstein-Barr virus is associated with NPC, genomic instability resulting in oncogenic fusion proteins may play an important role in the pathogenesis of this malignancy.³ While fusion proteins are poorly characterized in NPC, some have been identified that raise diagnostic and therapeutic opportunities. A fusion involving ubiquitin protein ligase E3 component n-recognin 5 (UBR5) and zinc finger protein 423 (ZNF423) has been recurrently detected in EBV-positive NPC tumor lines. Remarkably, UBR5-ZNF423 fusion protein expression is a requisite for the growth of cells harboring this rearrangement.⁴¹ The exact mechanism by which it drives tumorigenesis has not been described in detail, though is thought to involve incapacitation of EBF3 (early B cell factor), a tumor suppressor consistently expressed in NPC tumors.⁴¹

More recently, Yuan et al detected recurrent FGFR3-TACC fusion transcripts in NPC. This fusion has also been observed in head and neck cancer and esophageal squamous cell carcinoma.³ FGFR3 and TACC both reside on chromosome 4, and their intra-chromosomal fusion results in enhanced cell proliferation, colony formation, and transforming ability dependent on kinase activity. The central role that kinase activity plays in the oncogenic pathway of FGFR3-TACC positive NPC is demonstrated by increased phosphorylation of ERK and Atk, suggesting activation of these signaling pathways, and by the significant decrease in colony formation and growth rate of kinase-dead FGFR3-TACC 508M mutants and FGFR3-TACC positive cells treated with an FGFR3 kinase inhibitor.³ The dose-dependent attenuation of cell proliferation rate upon administration of a kinase inhibitor marks FGFR3-TACC as a druggable fusion, thereby raising a therapeutic opportunity for the treatment of patients with fusion positive NPC.

Fusion positive CAMSG has also been identified in the nasopharynx. An investigation examining PRKD fusions in PALG and CAMSG studied 3 tumor samples arising from the nasopharynx. All three tumors were positive for a PRKD1 fusion.¹⁰

Midline head and neck tumors

A poorly differentiated aggressive cancer with a predilection for midline structures of the head and neck is characterized by the BRD4-NUT fusion resulting from a t(15;19)(q13;p13) translocation.^8,42 This tumor tends to afflict children and young adults, although it has been described in older patients.^8,42 BRD4 resides on chromosome 19p3 and encodes for a chromatin interacting nuclear domain. Its function is thought to involve the preservation of cellular memory throughout cell division by activating genes prior to mitosis and restarting their transcription after mitosis via the recruitment of the transcription elongation complex P-TEFb. NUT is found on chromosome 15q3 and encodes a poorly understood protein mainly expressed in the testes. A chromosomal breakpoint results in the fusion of the 5′ end of BRD4 to nearly the entire sequence of NUT. Alternative BRD3-NUT fusions resulting from t(9;15)(q34;q13) have also been identified. Both fusions appear to interfere with epithelial differentiation and promote tumor growth.⁸

T(15;19) positive tumors are considered to be invariably fatal, however one case report describes a patient with this type of midline tumor who was cured with multimodality therapy involving hyperfractionated accelerated radiotherapy and 4 chemotherapy cycles.⁴² After treatment, biopsy revealed no viable tumor cells, and the patient had been in remission for 13 years, challenging the notion that tumors harboring the BRD4-NUT fusion are incurable.⁴²

Cytologic/molecular diagnostics

Fine needle aspiration biopsy (FNAB) is still the initial diagnostic method of choice for many salivary gland neoplasms. Current success of FNAB in salivary gland neoplasm diagnosis is largely because of its high sensitivity and specificity in: 1) distinguishing neoplastic from non-neoplastic salivary gland lesions and 2) distinguishing benign and low grade salivary gland neoplasm from high grade malignancies. However, precise subtyping of salivary gland neoplasms remains a diagnostic problem for some cases of FNAB. These challenging cases include: 1) low grade malignancies with minimal malignant cytological features; 2) accurate subtyping of high grade malignancies, including metastatic malignancies. Emerging molecular markers including specific chromosomal translocations provide real hope to improve the diagnostic accuracy of salivary gland FNAB. Compared to classic immunhistochemical markers, genetic changes can be highly specific for a particular subtype of salivary gland neoplasms.

A series of recent studies have made initial efforts to translate the newly identified markers into more accurate diagnosis of routine salivary gland FNABs. Immunohistochemical staining is an integral part of modern pathological diagnosis worldwide. Commercially available antibodies to various fusion proteins, products of gene rearrangements, are becoming increasingly available. Pusztaszeri et al and Foo et al both showed antibodies to MYB, PLAG1 and HMGA2 are useful in distinguishing ACCs from PAs.^43,44 Interestingly, a subset of ACCs without detectable MYB related gene rearrangement also overexpressed MYB mRNA and protein. Certain microRNAs (miRs) including MIR-15a, miR-16 and miR-150 are suggested to take part in the upregulation of MYB. Similarly, overexpression of PLAG1 protein has been identified in a subset of PA without detectable PLAG1 gene rearrangement. FISH is a classic method for cytogenetic analysis. Actually, the original identification of Myb t(6;9) (q22-23; p23-24) in ACC was done in surgically excised neoplasm using break-apart FISH probes.⁴⁵ Using FISH and salivary gland FNAB smears, Hudson and Collins demonstrated that 50% of ACCs contained MYB gene rearrangement or trisomy, while no such rearrangement was identified in 13 PAs.⁴⁶ Griffith CC et al detected ETV6 gene rearrangements in the cell blocks of all 3 tested MASCs.⁴⁷ While FISH analysis is highly specific, this technology is not readily available in every pathology laboratory. The impressive power of RT-PCR to amplify gene copy number is very attractive for many cytopathologists and molecular pathologists, especially because the FNAB specimens frequently contain limited tumor cells.⁴⁸ However, the complex genetic changes including probably unknown fusion gene partners might hinder current application of this technique into routine practice. A recent RT-PCR based study using primers of all known PLAG1 and HMGA2 fusion gene partners identified chromosomal translocation in only 12/45 PAs.⁴⁹ On the contrary, all 45 PAs were immunohistochemically positive for PLAG1 protein. Next generation sequencing (NGS) can be successfully conducted by using as little as 5-10 nanograms of DNA, well within the range of DNAs obtained by FNAB specimen. In combination with anchored multiplex

PCR, NGS can detect gene rearrangement even without prior knowledge of the fusion partners and gene copy number changes. Pilot studies using targeted NGS have already identified novel genetic changes critical to salivary gland tumor classification and treatment.⁵⁰ With the reduced cost of NGS, its application in routine FNAB diagnosis will significantly improve the diagnostic accuracy.

NUT midline carcinomas also pose a diagnostic dilemma, with the need to be differentiated from poorly differentiated or undifferentiated carcinoma, squamous cell carcinoma, Ewing sarcoma, sinonasal undifferentiated carcinoma, thymic carcinoma, and neuroblastoma.⁵¹ This differentiation is important because NUT midline carcinomas have a much poorer prognosis than those of non-NUT midline carcinoma. Pathologically, diagnosis of NUT midline carcinoma is established by demonstration of NUT rearrangement by FISH or by demonstration of BRD4-NUT fusion transcript usually by RP-PCR. C52, a rabbit monocloncal antibody against a recombinant polypeptide, was recently produced. Haack et al validated this antibody using 1068 tissue preparations (both resection specimens and tissue microarrays).The sensitivity and specificity of C52 for NUT midline carcinoma diagnosis are 87% and 100%, respectively.⁵² Recently it has been reported that NUT midline carcinomas are responsive to bromodomain and extra-terminal inhibitors, whose therapeutic target is BRD4-NUT fusion protein.⁵³

Conclusion

Fusion proteins are becoming an increasingly relevant entity in head and neck cancers as more of them are identified in specific head and neck malignancies. Identification of fusions harbored by different malignancies helps to elucidate tumor pathogenesis, biological behavior, and prognosis, thus providing valuable opportunities for targeted interventions to treat head and neck cancers. Therefore, there is a pressing need to better characterize fusion proteins in head and neck malignancies.

Table 1.

Fusion genes in head and neck neoplasms.¹⁰

Neoplasm	Gene/Fusion Transcript	Role in Pathogenesis
Salivary gland neoplasms
CAMSG	ARID1A-PRKD1	Possible deficiency in DNA repair; possible aberrant signal transduction*
	DDX3X-PRKD1	Possible deficiency in DNA repair*
ACC	MYB-NFIB	MYB oncogene activation
MEC	MECT1†-MAML2	AREG-EGFR signaling upregulation
MASC	ETV6-NTRK3	Increased STAT5A transcription; PI3k-AKT and RAS-MAP pathway activation
PA	CTNNB1-PLAG1	PLAG1 transcriptional factor activation and protein overexpression
	HMGA2-WIF1	HMGA2 chromatin remodeling gene upregulation
HCCC	EWSR1-ATF1	Unknown
Thyroid neoplasms
PTC	STRN-ALK	Constitutive ALK kinase activation
	GTF2IRD1-ALK	Constitutive ALK kinase activation
	NCOA4-RET	Constitutive RET kinase activation
	CCDC6-RET	Constitutive RET kinase activation
	AKAP13-RET	Constitutive RET kinase activation
	FKP15-RET	Constitutive RET kinase activation
	SPECC1L-RET	Constitutive RET kinase activation
	TBLXR1-RET	Constitutive RET kinase activation
	AKAP9-BRAF	Constitutive BRAF kinase activation
	Z3HAV1-BRAF	Constitutive BRAF kinase activation
	FAM114A2-BRAF	Constitutive BRAF kinase activation
	MKRN1-BRAF	Constitutive BRAF kinase activation
	AGGF1-RAF1	Constitutive RAF1 kinase activation
	TFG-MET	Constitutive MET kinase activation
	ETV6-NTRK3	NTRK kinase activation; PI3k-AKT and RAS-MAP pathway activation
	HEPHL-PANX1	Unknown
	METTL1-FAM53B	Unknown
	KIAA126-ARL17B	Unknown
	PPIP5K1-CATSPER2	Unknown
	RHOBTB2-PEBP4	Unknown
FTC	PAX8-PPARG1	Wild-type PPARG1 downregulation
	CREB3L2-PPARG1	cAMP-responsive transcription inhibition
FVPTC	PAX8-PPARG1	Wild-type PPARG1 downregulation
MEC	MECT1-MAML2	AREG-EGFR signaling upregulation
Nasopharynx
NP	UBR5-ZNF423	EBF3 tumor suppressor inactivation
	FGFR3-TACC	Kinase dependent proliferation
Midline head and neck tumors
NMC	BRD4-NUT	Interference with epithelial differentiation
	BRD3-NUT	Interference with epithelial differentiation

CAMSG, cribiform adenocarcinoma of minor salivary gland origin; ACC, adenoid cystic carcinoma; MEC, mucoepidermoid carcinoma; MASC, mammary analog secretory carcinoma; PA, pleomorphic adenoma; HCC, hyalinizing clear cell carcinoma; PTC, papillary thyroid carcinoma; FTC, follicular thyroid carcinoma; FVTC, follicular variant papillary thyroid carcinoma, nasopharyngeal carcinoma; NMC, NUT midline carcinoma.

^*Exact functional role of fusion unclear, however, function of participating genes known (see text).

^†Also known as CRTC1.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Author contributions

C.F. responsible for manuscript conception and design. D.E.,H.W., and C.F. collected the data used. D.E. wrote original draft, including introduction and sections on salivary gland neoplasms, thyroid neoplasms, nasopharyngeal carcinoma, and midline tumors. H.W. contributed cytologic/molecular diagnostic section. C.F. reviewed and edited manuscript and contributed sections on MASC and HCCC. C.F. supervised planning and execution of overall project and approved final version of manuscript to be submitted D.E. and C.F. are responsible for the overall content as guarantors.