Genetic Profile of Adenomatoid Odontogenic Tumor and Ameloblastoma. A Systematic Review

Purpose: To perform a comprehensive and systematic critical appraisal of the genetic alterations reported to be present in adenomatoid odontogenic tumor (AOT) compared to ameloblastoma (AM), to aid in the understanding in their development and different behavior. Methods: An electronic search was conducted in PubMed, Scopus, and Web of Science during March 2021. Eligibility criteria included publications on humans which included genetic analysis of AOT or AM. Results: A total of 43 articles reporting 59 AOTs and 680 AMs were included. Different genomic techniques were used, including whole-exome sequencing, direct sequencing, targeted next-generation sequencing panels and TaqMan allele-specific qPCR. Somatic mutations affecting KRAS were identified in 75.9% of all AOTs, mainly G12V; whereas a 71% of the AMs harbored BRAF mutations, mainly V600E. Conclusions: The available genetic data reports that AOTs and AM harbor somatic mutations in well-known oncogenes, being KRAS G12V/R and BRAFV600E mutations the most common, respectively. The relatively high frequency of ameloblastoma compared to other odontogenic tumors, such as AOT, has facilitated the performance of different sequencing techniques, allowing the discovery of different mutational signatures. On the contrary, the low frequency of AOTs is an important limitation for this. The number of studies that have a assessed the genetic landscape of AOT is still very limited, not providing enough evidence to draw a conclusion regarding the relationship between the genomic alterations and its clinical behavior. Thus, the presence of other mutational signatures with clinical impact, co-occurring with background KRAS mutations or in wild-type KRAS cases, cannot be ruled out. Since BRAF and RAS are in the same MAPK pathway, it is interesting that ameloblastomas, frequently associated with BRAFV600E mutation have aggressive clinical behavior, but in contrast, AOTs, frequently associated with RAS mutations have indolent behavior. Functional studies might be required to solve this question.


INTRODUCTION
Adenomatoid Odontogenic Tumor (AOT) and Ameloblastoma (AM) are benign epithelial odontogenic tumors affecting most commonly the tooth bearing areas of the jaws. Both tumors are composed of a proliferation of epithelial cells arranged in a way that reminds to some extent, to the enamel organ of a tooth germ [1]. AM is well-known for being locally infiltrating, for its continuous growth, its high rates of recurrences if not adequately removed and the possibility of undergoing malignant transformation [2]. On the contrary, AOT manifests clinically as a slow and self-limiting growth which does not require the aggressive surgical approach usually adopted for AM. In AOT, recurrences are extremely rare, even if it is partially removed [3]. The clinicopathological features of both tumors, allows to consider AM as a neoplasm, whereas there is a general agreement that AOT may represent a hamartoma [3][4][5], although this is a matter of debate.
Little information is known about the genetic background of AOT. Mutations in the β-catenin gene (CTNNB1) have been suggested, as strong cytoplasmatic expression of β-catenin is reported using immunohistochemistry [4,21]. However, authors have failed to show alterations in CTNNB1 [4]. Nevertheless, other more recent studies have shown consistent mutations in KRAS [22][23][24] and copy number alterations [23] affecting IGF2BP3.
As both AM and AOT have different clinical behavior, suggesting a different biological nature, and there has been a significant interest in papers reporting their genetic alterations during the last years, the aim of this systematic review was to compare the genetic alterations of AOT with the ones reported in AM, in order to summarize the current genetic knowledge of these lesions and aid in the understanding of the genomic alterations underlying their development.

METHODS
This systematic review was conducted following the PRISMA Statement guidelines.

Eligibility Criteria
Parameters were kept broad to maximize search results. The inclusion criteria consisted of full text observational research studies on humans about genetic analysis of adenomatoid odontogenic tumor or ameloblastoma, with or without clinicopathological and treatment information. Studies were excluded if they were about polymorphisms, were performed in-vitro or were not performed on human participants. Conference abstracts, articles where the full text was unavailable, bioinformatic research only, reviews, case reports, or case series without genetic analysis were also excluded.

Information Sources and Search Strategy
A preliminary literature search was conducted by one of the authors (RMF) to guide the search strategy. The search was conducted in PubMed, Scopus, and Web of Science during March of 2021, restricted to human studies in English language and without year restrictions. The following keywords were used in the identification of potential articles: (adenomatoid odontogenic tumor OR ameloblastoma) AND (chromosomal alteration OR copy number variation OR deletion OR gene mutation OR genetic OR genomic OR genome OR insertion OR loss of heterozygosity OR microarray OR sanger sequencing OR single nucleotide variant OR targeted next-generation sequencing OR whole exome sequencing). This was further complemented with manual searches using the reference list of each identified study.

Selection Process
After the removal of duplicates, two independent researchers (SN and CM) read the title and abstract to identify and select articles. Full text of selected studies were then analyzed and those who met the eligibility criteria were included in the review (Figure 1). Any disagreement was resolved through discussion guided by a third researcher who acted as a referee (RMF). No automation tools were used in this process

Data Collection Process
After reading the full text, two independent researchers (SN and CM) extracted and transferred the data to a Microsoft Excel spreadsheet (Microsoft Office 365 R ). Any disagreement was resolved through discussion guided by a third researcher who acted as a referee (RMF). No automation tools were used in this process.

Study Risk of Bias Assessment
The entities included in this review are rare and as such, the highest quality of primary data is from case series. At present, there is no agreed guidelines to perform and report molecular biology studies about odontogenic tumors. Hence, there is substantial heterogeneity in their data recording and reporting. Given these limitations, the risk of bias will be uncertain on almost all reported case series, with low quality of evidence. Therefore, we have decided not to undertake these assessments.

Synthesis Methods
A narrative synthesis of the data is planned. The characteristics collected from the studies to do the quantitative analysis will be based on: first author, year, country, tissue sample, sample size, gene or chromosome involved, gene mutation, signaling pathway, and genetic assay. Visualization of the data will be presented in form of figures and tables.

RESULTS
The search results are outlined in a PRISMA flow diagram in Figure 1. The initial literature search identified 677 studies. Following duplicate removal (n = 396), 281 studies had their titles and abstracts screened by two of the reviewers (SN and CM) from which 205 articles were removed. Finally, 76 studies were included for full text evaluation to ensure they satisfied the inclusion and exclusion criteria. Thirty-four articles were excluded with the following reasons: reviews (n = 14) and wrong study design (n = 19). In total, 43 articles were included in this systematic review (Figure 1).

Gene Mutations
Six articles reported mutations in AOT [4,[22][23][24][25][26]. A total of 59 AOTs were assessed under different genomic techniques, such as direct sequencing, targeted NGS panels, and TaqMan allele-specific qPCR ( Table 1). KRAS was the most commonly affected gene among the studies that included this gene in their analysis. One article used TaqMan allele-specific qPCR for KRAS [22], whereas other two worked with a Targeted NGS panel which included RAS family [23,24] (Table 1). A total of 54 samples were analyzed among these 3 studies, from which 75.9% harbored somatic mutations in KRAS (n = 41). All the studies reported that the mutations affecting KRAS corresponded to single nucleotide variations corresponding to missense mutations [22][23][24]. All the mutations affected codon 12, in which three types of transversions were identified: guanine to thymidine (G>T) in 24/41 cases, guanine to cytosine (G>C) in 16/41 cases and guanine to an adenosine (G>A) in one case. The aforementioned single base variations lead to G12V, G12R, or G12D substitutions, respectively [22][23][24] (Figures 2, 3).
Among the remaining three studies, one article worked with a Targeted NGS panel assessing specifically mutations in SMO, BRAF, PTCH1, and GNAS [25]. Only one AOT was included and showed two missense mutations in SMO (Y394S and p.Y399S). No mutations affecting BRAF or PTCH1 were reported [25].  By direct sequencing, one study identified one heterozygous missense mutation affecting AMBN, leading to a R90W substitution [26]. Using pyrosequencing and direct sequencing, Harnet et al., did not detect mutations affecting CTNNB1 [4] in a follicular type of AOT ( Table 1).

Chromosomal Alterations
To date there is only one published article about chromosomal alterations in AOT [23]. By using a whole-genome array and comparing those results with databases, Gomes et al. reported two rare losses in AOTs. One deletion affecting IGF2BP3 at 7p15.3 in a single AOT, and another affecting the chromosome 6 at q15, but no gene was identified at that position. The deletion in IGF2BP3 involved an intronic region of the protein-coding transcript, however, in silico analysis predicted the implication of the first exon of four alternative transcripts. Its potential in tumorigenesis remains unclear [23] ( Table 2).
More than 25 different mutations were identified in ameloblastoma. BRAF was the most frequently affected gene. BRAF mutations were assessed in 23 studies, involving a total of 530 tumors. Approximately, 71% (n = 377) of the analyzed tumors harbored somatic mutations in BRAF being the V600E mutation the most commonly reported (Figure 3). Other singlenucleotide transversions affecting BRAF were also reported but were uncommon; BRAF T440P was found in two cases [25] and BRAF L597R in one case [6].
SMO was assessed in nine studies [6, 8-10, 13-16, 25] in which 264 tumors were evaluated trough Sanger sequencing, targeted NGS panels and WES. Somatic point mutations in SMO were present in 10.6% (n = 28) of the analyzed samples. L412F was the most common mutation found in 15 cases, followed by W535L in 4 cases and G416E in only one case. Ras activates BRAF which facilitates the phosphorylation of MEK, which in turn allows the phosphorylation and activation of ERK. The resulting signaling cascade culminates with translocation of ERK to the nucleus, and the activation of transcription factors that result in the expression of genes related to proliferation, differentiation, and survival. In (B,C), in the presence of oncogenic BRAF and KRAS, respectively, the constitutive activation is independent of extracellular factors and does not respond to biochemical signals that would normally regulate the activity. Adapted from "Vemurafenib in Oncogenic BRAF Signaling Pathway in Melanoma," by BioRender.com (2021). Retrieved from: https://app.Biorender.com/biorender-templates.
Mutations in the tumor suppressor gene TP53, were of low frequency reported only by two studies. Shibata et al. found TP53 mutations in 1 of 12 ameloblastomas [49] and Bartels et al. in 1 of 7 [12]. Kumamoto et al., were unable to find TP53 mutations in their cohort of 10 ameloblastomas [47]. Mutations in other tumor suppressor genes, such as PTEN, are also of low frequency and have been reported in 5/20 [34], 1/7 (12), and 2/62 ameloblastomas [10].
One article reported 45% of their cohort (9/20 ameloblastomas) to harbor missense mutations (in non-SNP sites) affecting TSC1. Correspondingly, those samples showed significantly lower mRNA expression levels compared to normal mucosa, suggesting a higher proliferation rate in ameloblastoma attributed to abnormal mTOR accumulation [40].
Two articles that performed WES reported the presence of mutations affecting KMT2D occurring in the background of BRAF mutations [13,15]. Guan et al. [13], reported 2/10 ameloblastomas to harbor non-sense mutations in KMT2D, whereas Shi et al. [15], identified 1/4 ameloblastomas with a frameshift deletion in the same gene. Odontogenesis-related genes have been widely associated with the etiopathogenesis of ameloblastoma. Somatic mutations in BCOR (inactivating frameshift deletion) LRP6, SCN5A (missense mutations in both), and LAMB1 (frameshift deletion) were identified with WES [13], and co-occurred in the background of BRAF mutations. Three missense and one splicing site mutations affecting the ameloblastin gene (AMBN) were found in 4/4 ameloblastomas by direct sequencing [26]. Mutations related to Wnt/β-catenin pathway have been reported affecting CTNNB1 in 3.3% (9/272) of the cases [8,10,13,16,42,44,45,48]. Likewise, another member of this pathway, APC, was reported by one study to be mutated in 3/6 cases [42] and another study reported four different single nucleotide variations affecting different locus at this gene in a cohort of 30 patients, with a mutation rate that ranged from 6.25 to 27.5% [37] (Table 3). Contrary to this, Tanahashi et al., assessed CTNNB1, APC, AXIN1, and AXIN2 in 18 ameloblastomas, and did not identify any missense mutations in these genes. However, the authors found one silent mutation in AXIN1 and one single nucleotide polymorphism (SNP) in AXIN2 [43]. Similarly, Siriwardena et al., did not identify mutations in CTNNB1 among six ameloblastomas [42].

DISCUSSION
Odontogenic tumors (OT) arise from dental tissues or their remnants, and for decades, this statement was only supported by the histologic appearance of these lesions, which resembles the enamel organ, dental papilla or the dental follicle [1,50,51]. Increasing evidence showing mutations and/or chromosomal alterations in the same genes expressed during odontogenesis have confirmed this association, consolidating the close relationship between ontogenesis and oncogenesis [8,9,14,16,38,[52][53][54][55][56][57][58][59][60]. The ongoing development of the   molecular aspects of odontogenic tumors has revolutionized the understanding of the ethiopathogenesis of these heterogenous group of lesions, allowing the proposal of novel molecular therapeutic targets. However, the exact mechanism underlying the tumorigenic process and possible causes for their different clinical behavior remains unknown. Both AOT and AM are odontogenic tumors of epithelial origin, but their clinical behavior is diametrically different, resulting in different treatment approaches and prognosis. In the current systematic review, we aimed to compare the genetic alterations of AOT with the ones reported in AM, in order to summarize the current genetic knowledge of these lesions and aid in the understanding of the genomic alterations underlying their development and different behavior.
Our search identified six studies that analyzed the genetic aspects of AOTs (n = 59), in contrast to 37 that explored the genetic landscape of ameloblastoma (n = 530). Mutation in exon 12 of KRAS was found to be present in 76% of AOTs with G12V/R being the most found [22,23]. On the other hand, mutations in BRAF were found in 71.1% of the samples, corresponding mainly to V600E [6,25] (Figure 3). Interestingly, the proportion of AOTs with KRAS driver mutation, is similar to the proportion of driver mutations reported in ameloblastoma. Due to its high frequency, KRAS mutations were proposed as a driver mutation and signature marker of AOTs [22][23][24]. Because KRAS mutations are a recurrent finding in AOTs, Coura et al. [22], suggested the presence of KRAS G12V/R to help in the diagnosis of controversial cases of AOT, in the same way that BRAF V600E could be used in routine ameloblastoma diagnostics [14,38].
The RAS oncogene family is comprised of three members, KRAS, NRAS, and HRAS, and plays an important role in normal development, but also for cancer development. Activated point mutations on RAS proteins are widely present across a different spectrum of human cancers [61,62]. Our review showed that all KRAS mutations reported in AOTs have been found affecting codon 12 [22][23][24]. Mutations affecting this codon have been reported in non-small cell lung cancer and pancreatic ductal adenocarcinoma, being present in almost half of the cases for the former, and in 16% for the latter [63,64]. KRAS corresponds to a small GTPase that transduces extracellular signals to intracellular signal transduction cascades [65] (Figure 3). It has been suggested that the mutation subtype may affect downstream signaling differently, which could be reflected clinically [66,67]. Nevertheless, to date, this has not been demonstrated in AOTs. Coura et al., reported in their cohort of 38 AOTs, no statistically significant association between the presence of mutations (mainly KRAS G12V and G12R) and clinicopathological parameters (including patient's age, tumor size, location, follicular or extrafollicular variants, and fibrous capsule thickness) [22].
Recently, Bello et al. [68] proposed that the interactions between the adhesion proteins FAK, paxillin and PI3K may be relevant in the aggressiveness of AM compared to AOT, based on the observation that FAK expression was stronger in AM compared to AOT, and that one case of peripheral AM with strong expression of the three proteins had a history of two recurrences. Nevertheless, their conclusions should be carefully interpreted because were based on observations based on a small cohort of AOTs (n = 7).
The biological nature of AOT has been a constant matter of debate. In 2017, Reichart et al. [69], compared the immunohistochemical expression of different factors between AOT and AM, and proposed AOT to be a hamartomatous process rather than a true neoplasm [69]. Markers related to invasion, such as cytokeratin profiles and integrins, to proliferation, such as MDM2, p53 protein and metallothionein levels, were found to be higher in ameloblastomas compared to AOTs. Also, AOTs showed lower levels of matrix metalloproteinases (consistent with a reduced local aggressiveness), Ki76 and anti-apoptosis markers such as Bcl-2, and higher levels of β-catenin (suggesting greater cell adhesion properties) [69]. Similarly, there are publications about the strong cytoplasmatic expression of β-catenin on AOTs [4,21], however no mutation in CTNNB1 was detected [4]. The proposal of Reichart et al. [69] was based purely on immunohistochemical findings without considering genetic aspect. Although our knowledge about the molecular background of AOT is still very limited, the genetic data collected by this review points to the direction that AOT harbor mutations in important oncogenic driver genes, such as KRAS, and based purely on this, some authors have proposed it as a neoplasm [22]. Nevertheless, until now, the presence of these genetic alteration seems not to have a direct impact on its clinical behavior. Thus, care has to be taken when interpreting these findings.
The low number of studies that have performed small to large-scale and/or "omics" techniques to characterize the molecular background of AOTs, the low frequency of AOT (accounting for <5% of odontogenic tumors) [70][71][72], limited clinical information availability, and the fact that most of the available studies come from single-institution series or case reports, limit the conclusions than can be drawed out of these findings. Also, current publications are all retrospective studies based on formalin-fixed paraffin-embedded samples (much of them subjected to decalcifications methods), which shows inherent limitations, mainly related to the quality of the nucleic acids for these purposes and the difficulty of retrieving a large cohort. Nevertheless, molecular pathology is demonstrating its utility in the diagnosis of challenging cases and for targeted therapy of disfiguring tumors such as ameloblastoma, avoiding considerable postsurgical morbidities.

CONCLUSIONS
The available genetic data reports that 75% of AOTs harbor somatic mutations in KRAS, a well-known oncogene. Nevertheless, the number of studies that have a assessed the genetic landscape of AOT is still very limited, not providing enough evidence to draw a conclusion regarding the relationship between the genomic alterations and its clinical behavior. There are a significant number of studies that have assessed the genetic aspects of ameloblastoma. Different genetic alterations have been reported, being the BRAFV600E mutation the most common. The relatively high frequency of ameloblastoma compared to other odontogenic tumors, such as AOT, has facilitated the performance of different sequencing techniques, allowing the discovery of different mutational signatures. On the contrary, the low frequency of AOTs is an important limitation for this. Thus, the presence of other mutational signatures with clinical impact, co-occurring with KRAS background or in wild-type KRAS cases, cannot be ruled out.
Since BRAF and RAS are in the same MAPK pathway, it is interesting that ameloblastomas, frequently associated with BRAFV600E mutation have aggressive clinical behavior, but in contrast, AOTs, frequently associated with RAS mutations have indolent behavior. Functional studies might be required to solve this question.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.