The landscape of genetic alterations in ameloblastomas relates to clinical features

Ameloblastoma is a mostly benign, locally invasive odontogenic neoplasm arising in the jaws. The tumor originates from the epithelium involved in tooth formation, the enamel organ, epithelial cell rests of Malassez, reduced enamel epithelium, and epithelial lining of odontogenic cysts with special reference to dentigerous cysts [1, 2].

Tooth development (odontogenesis) is being initiated by interactions between epithelial and mesenchymal cells derived from the ectoderm of the first branchial arch and the ectomesenchyme of the neural crest. Odontogenesis involves several morphologically distinct stages. The mesenchyme of the developing tooth induces epithelial proliferation forming the dental lamina at the sixth week of gestation. Small epithelial cell nests invade the underlying mesenchyme and are referred to as bud stage. This phase is followed by proliferation and condensation of the mesenchyme around the bud, which then invaginates to form a cap shape known as cap stage at gestational weeks 9 to 11. The enamel organ starts being formed in the cap stage, then further differentiates into an outer and inner enamel epithelium, and further forms the stellate reticulum, imparting a bell shape, known as the bell stage. Ameloblasts arise from the inner epithelium. Ameloblastic epithelium shows nuclear polarization toward the underlying reticulum surface, known as the stratum intermedium thought to assist enamel production, which is not present in ameloblastomas, and therefore, no enamel is being produced. Epithelial polarization is recapitulated in ameloblastic tumors and marks the formation of pre-ameloblasts, which together with odontoblasts from the dental papilla induce dentin and enamel [3].

Ameloblastomas represent approximately 1% of all oral tumors and about 9 to 11% of all odontogenic tumors. Approximately 80% of ameloblastomas arise in the mandible, foremost in the third molar region, and the remaining 20% in the maxilla [1, 2]. Ameloblastomas were classified by the World Health Organization (WHO) into solid/multicystic, peripheral (extraosseous counterpart of the intraosseous solid/multicystic ameloblastoma), desmoplastic, and unicystic types with implications for treatment [4]. However, in the new 2017 classification of WHO, ameloblastomas were narrowed to ameloblastoma (conventional), unicystic, extraosseous/peripheral, and metastasizing variants due to the introduction of prospective views based on updates from genetic studies [4].

Ameloblastomas typically present as locally aggressive odontogenic tumors, frequently asymptomatic and slow-growing, with no evidence of swelling. There is a high propensity for local recurrence if not adequately removed at the initial surgery and, though the tumor may appear microscopically benign, development of distant metastasis is possible. Ameloblastic tumors with cytological atypia are classified as ameloblastic carcinoma and have a propensity for rapid growth and metastasis. Rarely, ameloblastic neoplasms metastasize despite of a benign histologic appearance, and these tumors are classified as metastasizing ameloblastomas [1, 2]. Current treatment options for ameloblastomas include both conservative treatment (enucleation or curettage) and resection. The former is associated with high rates of recurrence, while the latter results in significant facial deformity and morbidity. Importantly, patients should be followed up life-long due to an unpredictable biological behavior, especially in the case of maxillary tumors which carry a worse overall prognosis as compared to mandibular ameloblastomas [1, 4].

Development of non-invasive therapies has been precluded by a lack of understanding the molecular pathology of ameloblastomas, and hence, no risk classification for the likelihood of recurrence is currently available. Low prevalence, frequently small tissue samples and degradation of DNA resulting from decalcification has not made ameloblastoma an easy candidate for molecular analyses. However, comprehensive and highly sensitive next-generation sequencing technologies paved the way. Recently, oncogenic mutations were discovered activating constitutively signal transduction pathways relating to developmental stages of odontogenesis including the mitogen-activated protein kinase (MAPK) and hedgehog pathways [5‐11].

These studies identified BRAF as the most frequently mutated gene causing constitutive activation of the MAPK pathway in mandibular ameloblastomas of younger age patients, whereas SMO mutations were identified to activate hedgehog signaling predominantly in maxillary ameloblastomas of older patients [6‐8]. Since in very low frequencies the two mutations coexist, it has been proposed that these genetic alterations define two etiologically independent ameloblastic entities [5, 8]. However, the emerging genetic dichotomy between the tumors of two anatomic locations and age with regard to BRAF and SMO is not yet fully elucidated. Moreover, it is clinically important to correlate mutational status and histological evaluation with clinical outcome in large multicenter case series [5]. To explore further such clinically relevant genotype-phenotype features, we here comprehensively analyzed a large series of ameloblastomas collected from the archives of three different centers from Germany, France, and Turkey.

Our data confirm previous studies who found that BRAF and SMO are by far the most frequent oncogenic driver mutations in ameloblastomas. These genetic alterations lead to constitutive activation of MAP kinase and hedgehog signaling pathways, respectively [6‐8].

We also confirm highly significant phenotypic differences in these two types of ameloblastomas as BRAF-mutant cases occurred preferentially in the mandible and at a much younger age (mean age 42 years) than SMO-mutant cases occurring preferentially in the maxilla at an older age (mean age 67 years). This data underscores an emerging appreciation of the anatomical specificity of driver mutations, which reflect distinctive odontogenic pathways in the upper and lower dentition [13]. There is evidence that the nature of the molecular signaling in the upper and lower jaws may vary. The dental formula is the same in both arches in mice and in humans, but the shape and morphologies of the homologous teeth in the two jaws are clearly distinct. Biochemical signaling differences have been demonstrated in the mouse for Dlx-1 and -22 [14] and also for the activin/follistatin genes [15]. Although it is not known how neural crest-derived cells migrating into the developing maxillary and mandibular regions develop the ability to respond differently to ectodermal signaling, reports of apparently independent genetic determination of maxillary and mandibular dentitions, based on tooth size data derived from twins, are consistent with the molecular evidence [16]. Assuming that mutations in transcriptionally active genes controlling expansion of ameloblasts occur in a stochastic manner, the genetic landscape of ameloblastomas provides additional evidence for the existence of distinct developmental cues during ameloblast expansion. In this context, it is interesting to note that we found multiple gene mutations only in European but not in Turkish patients. As there are no known environmental factors predisposing to ameloblastomas, it remains to be speculated that there are differences in genetic backgrounds defining different mutational spectra.

Interestingly, the hedgehog signaling pathway is also involved in pituitary formation during early vertebrate embryogenesis. Its activation is triggered by hedgehog ligand binding to a receptor complex formed by the transmembrane protein patched 1 (PTCH1). In the presence of the ligand, the frizzled class receptor, smoothened (SMO), is released from PTCH1 inhibition and activates the transcription factor gene family GLI1, GLI2, and GLI3 [17]. After the appearance of Rathke’s pouch, Sonic HH (SHH) expression is excluded from this region but remains in surrounding areas [16]. As different types of mutations in the hedgehog, the Wnt, and BRAF/MAPkinase pathways define distinct subtypes of craniopharyngiomas [18], there seems to be signaling analogy in the pathways driving dentition and Rathke’s pouch formation [19]. Craniopharyngiomas are generally considered to arise from the remnants of Rathke’s pouch or a misplaced enamel organ [19]. Gomes et al. [18] hypothesized that crosstalk between Wnt/β-catenin and SHH pathways, which are important during pituitary embryogenesis, could contribute to the imbalance in intracellular signaling in the molecular pathogenesis of adamantinomatous craniopharyngiomas. Taken all together, we may speculate that ameloblastomas and craniopharyngiomas share similar tumorigenic pathways and SMO-mutated ameloblastomas may resemble a distinct subtype of craniopharyngiomas with special reference to maxillary location.

In line with these molecular subtypes, we found a clear correlation between the risk of recurrence and the mutational status. The recurrence rate was highest in the group of ameloblastomas with multiple gene mutations. Tumors with BRAF mutations revealed significantly lower risk for recurrence, and tumors with SMO gene mutation appear to be associated with higher recurrence [8]. Thus, our data suggest that a stratified clinical management of ameloblastomas may be possible. Tumors with BRAF mutations are excellent candidates for neoadjuvant BRAF inhibitor treatment followed by limited surgical treatment in tumors with single BRAF mutations and extensive resection of tumors with BRAF and concurrent multiple mutations. In contrast SMO mutant, tumors require a priori definite surgical resection with wider margins as they carry a high risk of recurrence.

Our data also raise the possibility that there is a continuum from benign to locally recurrent ameloblastomas to ameloblastic carcinomas. While malignant ameloblastoma showing clear features of malignancy is generally accepted to be a different entity based on its ability to metastasize, it is possible that the histologically benign but “metastasizing ameloblastoma” might be one of the lesions accumulating several oncogenic mutations and thus, acquiring a higher potential for malignant growth. Therefore, a larger number of biologically malignant ameloblastomas need to be analyzed by deep sequencing in order to establish such a relationship.

Finally, we here describe a genotype-phenotype correlation as tumors with BRAF mutations were found in all three histological groups, whereas neither SMO nor multiple gene mutations were seen in unicystic type. Moreover, multiple gene mutations were exclusively present in solid ameloblastomas (100%). Peripheral type ameloblastomas revealed single somatic mutations in BRAF or SMO. Most follicular subtypes showed BRAF and multiple gene mutations, whereas most plexiform and mixed variants harbored either SMO, NRAS, HRAS, or EGFR mutations. Therefore, our results support the argument that the mutation status may be related to the histological pattern (follicular versus plexiform). In their study, Sweeney et al. [8] found that plexiform variants had a SMO mutation (p < 0.02), while most follicular and desmoplastic variants carried either SMO or BRAF mutation.

In summary, our data significantly extend previous studies and provide evidence that there are distinct molecular pathways driving ameloblastomas with different histological and clinical features possibly requiring different approaches for clinical management.