Comparative Analysis Between Dentinogenic Ghost Cell Tumor and Ghost Cell Odontogenic Carcinoma: A Systematic Review

Abstract

Dentinogenic ghost cell tumor (DGCT) and ghost cell odontogenic carcinoma (GCOC) form a spectrum of rare benign and malignant odontogenic neoplasms, respectively. The aim of this study was to perform a comparative systematic review of the clinicopathological, genetic, therapeutic, and prognostic features of DGCT and GCOC. The electronic search was performed until December 2020 on seven electronic databases. Case reports, series, and research studies with enough histopathological criteria for diagnosis and all genomic studies were included. Both DGCT and GCOC showed a male prevalence (p = 0.043), with mandibular and maxillary predilections, respectively (p = 0.008). Peripheral DGCT (DGCTp) affected most elderly people (p < 0.001), and central DGCT (DGCTc) and GCOC occurred mainly in younger individuals. Unilateral enlargement of maxilla or mandible was the most common clinical sign associated with a radiolucent or mixed image. Ameloblastomatous epithelium was often present in both neoplasms. Basaloid and large cells with vesicular nuclei were also frequently seen in GCOC. β-catenin expression and mutations (CTNNB1 gene) were found in DGCT and GCOC. Conservative surgery was mostly used for DGCTp, while radical resection was chosen for DGCTc and GCOC. High recurrence rates were found in DGCTc and GCOC. Metastasis occurred in 16.7% of GCOC cases and the 5-year survival rate was 72.6%. DGCT and GCOC share numerous clinicopathological features and demand a careful histopathological evaluation, considering the overlap features with other odontogenic tumors and the possibility of malignant transformation of DGCT. A strict regular post-operative follow-up is mandatory due to high recurrence rates and metastatic capacity in GCOC.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12105-021-01347-z.

Introduction

Dentinogenic ghost cell tumor (DGCT) and ghost cell odontogenic carcinoma (GCOC) form a spectrum of rare benign and malignant ghost cell odontogenic neoplasms (GCONs), responsible for approximately 4.92 and 2.46% of all ghost cell odontogenic lesions [1], and 0.38 and 0.23% of all odontogenic tumors [2], respectively. DGCT can occur as a central (DGCTc) or peripheral (DGCTp) lesion. The latter generally has a more indolent course [3–5]. GCOC can arise from a calcifying odontogenic cyst (COC), which is a cystic variant of the spectrum of the three ghost cell odontogenic lesions, from a DGCT or de novo, and can cause regional and distant metastases [1, 6, 7], The diagnosis of these neoplasms is based on the microscopical absence or presence of cytological atypia in the proliferative odontogenic epithelium with a variable amount of ghost cells, which may also present dentinoid material [8, 9]. Proliferation markers, as Ki-67, are usually used to support the malignant nature of the tumor, however, the lack of established cut-off points makes difficult its use as a gold standard criterion to differentiate GCOC from DGCT [6, 9–11].

The pathogenesis of GCONs is not yet fully elucidated. However, whether odontogenic or not, lesions presenting ghost cells appear to share similar developmental pathways involving β-catenin protein and mutations on its coding gene CTNNB1 [3p22.1], causing deregulation in the Wnt/β-catenin/TCF signaling pathway, related to various human tumorigenesis [12–17]. Recent studies have shown potential applications of target therapies in odontogenic tumors due to specific molecular alterations [18–21]. An updated review of the current knowledge about the genomic features of the GCONs can not only collaborate to clarify its etiology but also guide and foster future molecular studies that investigate new therapeutic proposals and the role of modulators of the Wnt/β-catenin/TCF pathway in human neoplasms [17].

A previous systematic review compared DGCT and GCOC [22], but only information about clinicoradiological data, treatment approaches, and recurrences was included. Thus this paper aims to perform a more comprehensive comparative systematic review of the demographic, clinical, imaging, histopathological, immunohistochemical, genomic, therapeutic, and prognostic features, in an attempt to clarify the pathogenesis, diagnosis, and outcomes involved in DGCT and GCOC.

Material and Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Statement guidelines (2020) [23].

Search Strategy

A systematic review was performed with publications until December 2020. A bibliographic search, without year or language restrictions, was conducted on seven databases: PubMed, LILACS (Latin American and Caribbean Health Sciences Literature), SciELO, Cochrane Collaboration Library, Scopus, Embase, and Web of Science. The search strategy used was: (“dentinogenic ghost cell tumor” OR “odontogenic ghost cell tumor” OR “calcifying ghost cell odontogenic tumor” OR “epithelial odontogenic ghost cell tumor” OR “ghost cell odontogenic carcinoma” OR “calcifying ghost cell odontogenic carcinoma” OR “malignant epithelial odontogenic ghost cell tumor” OR “malignant calcifying ghost cell odontogenic tumor” OR “malignant calcifying odontogenic cyst” OR “carcinoma arising in a calcifying odontogenic cyst”). Published articles that did not appear in the English language were translated for evaluation.

Inclusion and Exclusion Criteria

Inclusion criteria comprised case reports, series, and research studies that provided clinical, imaging, histopathological, immunohistochemical and/or genomic data of DGCT and GCOC. Full-text information of published cases was accessed for diagnostic confirmation, according to World Health Organization (WHO) [9]. The presence of increased mitotic activity and cytological atypia (hyperchromatism, nuclear/cellular pleomorphism, and atypical mitosis—with at least two features present) was mandatory for GCOC. High expression of Ki-67 associated with one of the two criteria previously mentioned also confirmed the diagnosis. Regional or distant metastasis was the only clinical feature used to prove a GCOC. Research articles for the immunohistochemical analysis should report the histopathological criteria or description of the sample. Exceptionally for genomic analysis, all publications with this information were included, due to the lack of histopathological characterization in this type of study and its relevance and scarcity in the literature.

Exclusion criteria were: (1) Case reports, series, and studies that lacked clinicopathological information to confirm the diagnosis (except for genetic studies); (2) Literature reviews, book chapters, congress posters, and epidemiological studies; (3) Unavailable full-text articles; (4) Hybrid/mixed odontogenic lesions; (5) Duplicated cases. We contacted the authors to identify possible duplicated cases that were not clear in the publications. The cases in which we did not have an answer from the authors were excluded.

Study Selection

The titles and abstracts of publications were independently assessed by G.S.V. and P.P.M. Articles that fit the selection criteria (clinicopathological and/or molecular information about the DGCT and GCOC) and revealed insufficient information in the abstract had the full text’s evaluated. Additional searches were done online and on the references of the selected articles to identify relevant studies not found in databases. Disagreements were solved with a discussion between authors and diagnostic confirmation was supported by experienced oral pathologists (D.C.C. and K.S.G.C.).

Data Extraction

Data to be extracted were collected independently by G.S.V. and P.P.M. For the demographical, clinical, imaging, therapeutic and prognostic comparison between DGCT and GCOC, the benign entity was also divided according to central and peripheral variants. For histopathological, immunohistochemical, and genomic comparison, the lesions were compared according to their benign and malignant nature. The following data were extracted of the selected articles, when available: country of publication, sex, race/ethnicity, age, location, evolution, signs and symptoms, clinical features of the lesion, imaging features, histopathological features, immunohistochemical labeling, genomic alterations, treatment approach, recurrence status, regional and/or distant metastasis (for GCOC), follow-up period, death from disease (for GCOC).

Quality Assessment and Data Analysis

Quality of the case reports and series was assessed using the Joana Briggs Institute (JBI) critical appraisal tools [24, 25]. According to our extracted data, the qualitative panorama was based on questions 1 to 6 of the case reports tool, and those of 6 to 8 for the case series tool. Mann–Whitney test was used to compare means of continuous variables according to their normality, which was evaluated using the Kolmogorov–Smirnov test. Effect sizes measured the strength of the relationship between two numerical variables in a sample-based estimate. Pearson Chi-Square and Fisher’s exact tests compared categorical variables frequencies according to the number of occurred events. Phi (ɸ) or Cramer’s V coefficients gauged the strength of a relationship (> 0.25 very strong and > 0.15 strong). The calculation of the odds ratio (OR) and 95% confidence interval (CI) estimated the risk between exposure and disease. Through the Kaplan–Meier method, the overall survival rates of GCOC were reached using survival analysis. Analyses were performed using IBM® SPSS Inc. software (version 20.0, Chicago, IL, USA), and statistical significance was assumed when p < 0.05.

Results

Study Selection

The flow diagram of the strategy results is shown in Fig. 1. The search found 1132 articles, and after removing duplicates, 732 articles had the title, abstract or complete text screened. After screening and inclusion of articles by reference and online search, 273 publications were considered for eligibility evaluation. According to the previously mentioned criteria, 135 articles were excluded, with two additional publications ruled out due to inconsistency between histopathological images and text descriptions. A total of 136 articles were included in the systematic review.

Fig. 1.

Systematic review flow diagram according to PRISMA guidelines

Quality Assessment

In total, 105 case reports were included, of which 14 (13.3%) obtained all positive responses in the 6 questions used as a parameter. Among the case series, none of the articles were positive in all the 3 questions. The detailed quality assessments are available in electronic supplementary materials 1 and 2.

Sample Description

Regarding the 136 included articles, published from 1962 to 2020, most publications were from Asia (58.1%)  – India (20.6%) and Japan (14.7%); followed by North America, Europe, South America, Africa, and collaborations between continents (electronic supplementary materials 3). In the demographic, clinical, imaging, histopathological, therapeutic and prognostic analysis, 130 DGCT, 76 (58.5%) DGCTc, 52 (40.0%) DGCTp and 2 (1.5%) DGCT without variant classification, and 48 GCOC were included. The immunohistochemical analysis included 49 DGCT and 36 GCOC, and in the genomic analysis, 7 DGCT and 6 GCOC.

Demographic, Clinical and Imaging Features

A male predilection was noted among benign and malignant GCONs (Pearson chi-square test, p value = 0.043; Φ = 0.152; OR 0.461; 95% CI 0.21–0.98). No statistical differences were observed between the average ages of DGCT and GCOC patients (Mann–Whitney test, p value = 0.468). However, DGCTp was more prevalent in elderly patients (54.04 ± 23.93) when compared to DGCTc (38.82 ± 19.40; Mann–Whitney test, p value < 0.001, Effect size: 0.392). A lack of racial/ethnic data was noted in 125 (70.2%) cases of GCONs included, hampering to obtain this panorama.

DGCT was more common in the mandible (61.5%) and GCOC occurred more frequently in the maxilla (62.5%; Fisher’s exact test, p value = 0.008; Cramer’s V = 0.234). The prevalence in mandible was maintained when we evaluated the DGCT variants separately. One (1,9%) DGCTp was reported in the ethmoid sinus. No statistical differences were noted between the average evolution' period of DGCT and GCOC (Mann–Whitney test, p value = 0.948). Enlargement associated with pain or not was the main clinical sign related to DGCT and GCOC. Associated teeth could present mobility and/or displacement. Maxillary and ethmoid sinus lesions were able to trigger sino-nasal symptoms. DGCTp had a smaller mean dimension than DGCTc (Mann–Whitney test, p value < 0.001, Effect size: 0.820), and GCOC was statistically larger than DGCT (Mann–Whitney test, p value < 0.001, Effect size: 0.533). Lymphadenopathy was associated with 10 (7.7%) DGCT and 5 (10.4%) GCOC.

DGCTp lacked imaging findings in 23 (44.2%) cases, and when present, the lesions were radiolucent or mixed and could cause erosion/saucerization (n = 16; 30.8%) of the underlying bone. DGCTc and GCOC were mainly mixed uni or multilocular images prone to cause bone cortex expansion (n = 36; 47.4%) and perforation (n = 28; 58.3%), respectively. Benign central lesions had more well-defined limits (n = 35; 46.1%) while malignant cases had usually ill-defined margins (n = 27; 56.3%; Fisher’s exact test, p < 0.001; Cramer’s V = 0.644). Tooth displacement, root resorption, impacted tooth, and infiltration to adjacent structures/bones were also observed in GCONs.

Demographic, clinical, and imaging features are detailed in Table 1.

Table 1.

Review of demographic, clinical, imaging, therapeutic and prognostic data obtained from published cases of DGCT and GCOC

	All DGCT n = 130a (100.0%)	Peripheral DGCT n = 52 (100.0%)	Central DGCT n = 76 (100.0%)	GCOC n = 48 (100.0%)
Demographic data
Sex
Male	79 (60.8%)	30 (57.7%)	47 (61.8%)	37 (77.1%)
Female	51 (39.2%)	22 (42.3%)	29 (38.2%)	11 (22.9%)
Race/ethnicity
Asian	11 (8.5%)	0 (0.0%)	10 (13.2%)	18 (37.5%)
Black	3 (2.3%)	3 (5.8%)	0 (0.0%)	4 (8.3%)
	11 (8.5%)	8 (15.4%)	3 (3.9%)	3 (6.3%)
White  Otherb	2 (1.5%)	0 (0.0%)	2 (2.6%)	1 (2.1%)
N.A.	103 (79.2%)	41 (78.8%)	61 (80.3%)	22 (45.8%)
Age (years)
0–10	6 (4.6%)	5 (9.6%)	1 (1.3%)	1 (2.1%)
11–20	18 (13.8%)	3 (5.8%)	15 (19.7%)	6 (12.5%)
21–30	15 (11.5%)	2 (3.8%)	13 (17.1%)	4 (8.3%)
31–40	19 (14.6%)	3 (5.8%)	14 (18.4%)	14 (29.2%)
41–50	18 (13.8%)	7 (13.5%)	11 (14.5%)	8 (16.7%)
51–60	11 (8.5%)	6 (11.5%)	5 (6.6%)	6 (12.5%)
61–70	19 (14.6%)	11 (21.2%)	8 (10.5%)	5 (10.4%)
71–80	18 (13.8%)	11 (21.2%)	7 (9.2%)	1 (2.1%)
81–90	3 (2.3%)	3 (5.8%)	0 (0.0%)	2 (4.2%)
91–100	1 (0.8%)	1 (1.9%)	0 (0.0%)	0 (0.0%)
N.A.	2 (1.5%)	0 (0.0%)	2 (2.6%)	1 (2.1%)
Mean ± SD; Min–Max	44.97 ± 22.43; 3–92	54.04 ± 23.93; 3–92	38.82 ± 19.40; 9–80	42.45 ± 18.55; 10–89
Clinical data
Evolution in months (Mean ± SD; Min–Max)	16.23 ± 24.83; 0.6–120	12.48 ± 16.59; 1–72	19.33 ± 29.69; 0.6–120	23.13 ± 33.38; 0.75–132
Sitec
Ethmoid sinus	1 (0.8%)	1 (1.9%)	0 (0.0%)	0 (0.0%)
Maxilla	47 (36.2%)	16 (30.8%)	29 (38.2%)	30 (62.5%)
Anterior	17 (13.1%)	12 (23.1%)	4 (5.3%)	1 (2.1%)
Posterior	10 (7.7%)	1 (1.9%)	8 (10.5%)	6 (12.5%)
Antero and posterior	14 (10.8%)	2 (3.8%)	12 (15.8%)	17 (35.4%)
N.A. maxillary location	6 (4.6%)	1 (1.9%)	5 (6.6%)	6 (12.5%)
Mandible	80 (61.5%)	34 (65.4%)	46 (60.5%)	18 (37.5%)
Anterior	14 (10.8%)	8 (15.4%)	6 (7.9%)	1 (2.1%)
Posterior	29 (22.3%)	13 (25.0%)	16 (21.1%)	6 (12.5%)
Antero and posterior	18 (13.8%)	5 (9.6%)	13 (17.1%)	9 (18.8%)
N.A. mandible location	19 (14.6%)	8 (15.4%)	11 (14.6%)	2 (4.2%)
N.A.	2 (1.5%)	1 (1.9%)	1 (1.3%)	0 (0.0%)
Cross midline (A.I.)	16 (12.3%)	1 (1.9%)	15 (19.7%)	12 (25.0%)
Signs and symptoms
Enlargement	96 (73.8%)	36 (69.2%)	58 (76.3%)	38 (79.2%)
Plaque lesion	2 (1.5%)	2 (3.8%)	0 (0.0%)	0 (0.0%)
Papillomatous lesion	1 (0.8%)	1 (1.9%)	0 (0.0%)	0 (0.0%)
Pain	28 (21.5%)	3 (5.8%)	25 (32.9%)	14 (29.2%)
Ulceration of the underlying mucosa and/or adjacent skin	7 (5.4%)	5 (9.6%)	2 (2.6%)	9 (18.8%)
Obliteration of the buccal sulcus	13 (10.0%)	0 (0.0%)	12 (15.8%)	3 (6.3%)
Paresthesia/numbness	1 (0.8%)	0 (0.0%)	1 (1.3%)	4 (8.3%)
Dysphagia	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
Reduced mouth opening (trismus)	2 (1.5%)	0 (0.0%)	2 (2.6%)	0 (0.0%)
Bleeding/epistaxis	0 (0.0%)	0 (0.0%)	0 (0.0%)	6 (12.5%)
Purulent discharge/secretion	4 (3.1%)	1 (1.9%)	3 (3.9%)	1 (2.1%)
Tooth mobility	10 (7.7%)	0 (0.0%)	9 (11.8%)	6 (12.5%)
Tooth displacement	7 (5.4%)	0 (0.0%)	6 (7.9%)	4 (8.3%)
Nasal obstruction/congestion	2 (1.5%)	1 (1.9%)	1 (1.3%)	3 (6.3%)
Sinus congestion	1 (0.8%)	0 (0.0%)	1 (1.3%)	0 (0.0%)
Post-nasal drip	1 (0.8%)	0 (0.0%)	1 (1.3%)	0 (0.0%)
Blepharoptosis	1 (0.8%)	0 (0.0%)	1 (1.3%)	1 (2.1%)
No alterations	1 (0.8%)	0 (0.0%)	1 (1.3%)	0 (0.0%)
N.A.	27 (20.8%)	11 (21.2%)	16 (21.1%)	9 (18.8%)
Consistency
Hard/firm	49 (37.7%)	19 (36.5%)	28 (36.8%)	7 (14.6%)
Soft	8 (6.2%)	5 (9.6%)	3 (3.9%)	0 (0.0%)
Fluctuant	2 (1.5%)	0 (0.0%)	2 (2.6%)	3 (6.3%)
N.A	71 (54.6%)	28 (53.8%)	43 (56.6%)	38 (79.2%)
Clinical dimension (centimeters)
Up to 3.0	51 (39.2%)	35 (67.3%)	15 (19.7%)	7 (14.6%)
3.1 to 6.0	21 (16.2%)	0 (0.0%)	20 (26.3%)	14 (29.2%)
> 6.0	10 (7.7%)	0 (0.0%)	10 (13.2%)	12 (25.0%)
N.A	48 (36.9%)	17 (32.7%)	31 (40.8%)	15 (31.3%)
Mean ± SD; Min–Max	3.29 ± 2.98; 0.3–20	1.46 ± 0.77; 0.3–3.0	4.78 ± 3.28; 1–20	5.97 ± 3.69; 0.3–21
Previous tooth extraction/traumatic avulsion (A.I.)	8 (6.2%)	2 (3.8%)	6 (7.9%)	3 (6.3%)
Edentulous/removable prosthesis site (A.I.)	15 (11.5%)	13 (25.0%)	1 (1.3%)	2 (4.2%)
Associated lymphadenopathy (A.I.)	10 (7.7%)	2 (3.8%)	8 (10.5%)	5 (10.4%)
Imaging data
No radiographic alterations	23 (17.7%)	23 (44.%)	–	–
Density
Mixed	47 (36.2%)	5 (9.6%)	42 (55.3%)	23 (47.9%)
Radiolucent/hypodense	34 (26.2%)	8 (15.4%)	26 (34.2%)	15 (31.3%)
N.A.	26 (20.0%)	16 (30.8%)	8 (10.5%)	10 (20.8%)
Locularity
Unilocular	30 (23.1%)	–	30 (39.5%)	11 (22.9%)
Multilocular	18 (13.8%)	–	19 (25.0%)	7 (14.6%)
N.A.	30 (23.1%)	–	27 (35.5%)	30 (62.5%)
Borders
Well-defined	35 (26.9%)	–	35 (46.1%)	5 (10.4%)
Partially well-defined	10 (7.7%)	–	10 (13.2%)	3 (6.3%)
Ill-defined	7 (5.4%)	–	7 (9.2%)	27 (56.3%)
N.A.	26 (20.0%)	–	24 (31.6%)	13 (27.1%)
Imaging findings
Bone expansion	36 (27.7%)	0 (0.0%)	36 (47.4%)	8 (16.7%)
Bone perforation	30 (23.1%)	0 (0.0%)	31 (40.8%)	28 (58.3%)
Bone erosion/saucerization	16 (12.3%)	16 (30.8%)	0 (0.0%)	2 (4.2%)
Impacted tooth	11 (8.5%)	0 (0.0%)	11 (14.5%)	3 (6.3%)
Tooth displacement	15 (11.5%)	1 (1.9%)	14 (18.4%)	11 (22.9%)
Root resorption	18 (13.8%)	1 (1.9%)	17 (22.4%)	12 (25.0%)
Infiltration to adjacent bones and/or maxillary sinus	7 (5.4%)	0 (0.0%)	7 (9.2%)	7 (14.6%)
No alterations	23 (17.7%)	23 (44.2%)	0 (0.0%)	–
N.A.	39 (30.0%)	12 (23.1%)	25 (32.9%)	8 (16.7%)
Therapeutic data
Conservative surgery	50 (38.5%)	28 (53.8%)	21 (27.6%)	1 (2.1%)
Conservative surgery + Radiotherapy	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
Radical surgery	42 (32.3%)	4 (7.7%)	37 (48.7%)	22 (45.8%)
Radical surgery + Neck dissection	0 (0.0%)	0 (0.0%)	0 (0.0%)	4 (8.3%)
Radical surgery + Radiotherapy	1 (0.8%)	0 (0.0%)	1 (1.3%)	9 (18.8%)
Radical surgery + Neck dissection + Radiotherapy	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
Radical surgery + Radiotherapy + Chemotherapy	1 (0.8%)	0 (0.0%)	1 (1.3%)	2 (4.2%)
Radical surgery + Neck Dissection + Radiotherapy + Chemotherapy	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
Radical surgery + Neck dissection + Radiotherapy + Chemotherapy + Immunotherapy	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
Patient refused treatment	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.1%)
N.A.	36 (27.7%)	20 (38.5%)	16 (21.1%)	5 (10.4%)
Prognostic data
Recurrence (months)
Yes	24 (18.5%)	1 (1.9%)	23 (30.3%)	19 (39.6%)
Time (Mean ± SD; Min–Max)	50.73 ± 66.71; 1–240	1.00; 1	53.10 ± 67.41; 3–240	24.41 ± 37.06; 0.63–144
No	59 (45.4%)	28 (53.8%)	30 (39.5%)	21 (43.8%)
N.A.	47 (36.2%)	23 (44.2%)	23 (30.3%)	8 (16.7%)
Metastasis
Yes	–	–	–	8 (16.7%)
Regional	–	–	–	2 (4.2%)
Distant	–	–	–	7 (14.6%)
Lung/pleura	–	–	–	5 (10.4%)
Brain/cranial	–	–	–	2 (4.2%)
Skeletal	–	–	–	1 (2.1%)
Cutaneous	–	–	–	1 (2.1%)
No	–	–	–	17 (35.4%)
N.A.	–	–	–	23 (47.9%)
Death from disease	1 (0.8%)	0 (0.0%)	1 (1.3%)	8 (16.7%)
Follow-up time in months (Mean ± SD; Min–Max)	40.85 ± 59.57; 3–372	32.15 ± 34.80; 6–156	45.67 ± 69.83; 3–372	42.85 ± 64.90; 2–336

SD standard deviation, A.I. available Information, N.A. not available

^aTwo cases included in this sample did not inform or did not provide enough information to categorize into peripheral or centra

^bRefers to single individuals designated as Hispanic, Guatemalan and Afro-Caribbean

^cAnterior region was defined in this review as incisor and canine regions, while the posterior premolar and molar regions

Histopathological Features

Ameloblastomatous parenchymal pattern was the most found in DGCT (n = 104; 80.0%) and GCOC (n = 27; 56.3%). However, basaloid (n = 26; 54.1%) and large cells with vesicular nuclei (n = 20; 41.7%), were also often related to cytological atypia in malignancies. Cribriform structures, spindle, epidermoid and clear cell differentiations were noted in GCONs. Dentinoid material was more common in benign lesions (n = 115; 88.5%) comparing to GCOC (n = 20; 41.7%), as well as calcifications (n = 72; 55.4% and n = 20;41.7%, respectively). Four (3.1%) DGCT were associated with inconspicuous mitotic figures and hyperchromatic/pleomorphic cells, not relevant for the diagnosis, and one (0.8%) case presented these features focally, suggesting a localized malignant transformation. Two (4.2%) GCOC with no cytological malignant aspects had metastasis and another two (4.2%) with only increased mitotic figures or cytological atypia and high Ki-67 proliferation index (PI) (between 49 and 60%). Infiltrative pattern (n = 20; 41.7%), presence of necrosis (n = 21; 43.8%), perineural (n = 3; 6.3%) and vascular invasion (n = 3; 6.3%) were important malignant features. Foreign body reactions to ghost cells were noticed in the stroma of GCONs.

GCOC arose de novo in 22 (45.8%) cases, and were preceded by the COC in 12 (25.0%) cases and by the DGCT in four (8.3%) cases. From the definitive treatment of precursors to GCOC diagnosis, the average period for the malignant transformation was 75.15 (± 71.56) months.

Histopathological features are detailed in Table 2.

Table 2.

Review of histopathological data obtained from published cases of DGCT and GCOC

	DGCT n = 130 (100.0%)	GCOC n = 48 (100.0%)
Histopathological data
No epithelial characterization	20 (15.4%)	5 (10.4%)
Epithelial characterization	110 (84.6%)	43 (89.6%)
Ameloblastomatous proliferation	104 (80.0%)	27 (56.3%)
Adenoid-like/cribriform aspect/pseudoductal structures	5 (3.8%)	6 (12.5%)
Basaloid cells	4 (3.1%)	26 (54.2%)
Large cells with vesicular nuclei	0 (0.0%)	20 (41.7%)
Spindle cells	5 (3.8%)	3 (6.3%)
Squamous/epidermoid cells	7 (5.4%)	7 (14.6%)
Clear cells	2 (1.5%)	10 (20.8%)
Mitotic figures	5 (3.8%)	45 (93.8%)
Atypical mitosis	–	9 (18.8%)
Cytological atypia (Hyperchromatism, nuclear/cellular pleomorphism and/or atypical mitosis)	5 (3.8%)	46 (95.8%)
Ghost-cells	130 (100.0%)	48 (100.0%)
Calcification	72 (55.4%)	20 (41.7%)
Dentinoid/osteodentin-like material	115 (88.5%)	20 (41.7%)
Multinucleated giant cells (Foreign body reaction to ghost cells located in the stroma)	9 (6.9%)	10 (20.8%)
Cystic/microcystic/pseudo-cystic spaces	38 (29.2%)	21 (43.8%)
Infiltrative/invasive front	11 (8.5%)	20 (41.7%)
Necrosis	–	21 (43.8%)
Perineural invasion	–	3 (6.3%)
Vascular invasion	–	3 (6.3%)
Precursor lesion
COC	–	12 (25.0%)
DGCT	–	4 (8.3%)
De novo	–	22 (45.8%)
Uncertain	–	4 (8.3%)
Non-precursor benign odontogenic tumors previously diagnosed at the same site	–	3 (6.3%)
N.A.	–	3 (6.3%)
Time for malignancy from precursor lesions (months) (Median ± SD; Min–Max)	–	75.15 ± 71.56; 7–228

N.A. not available, SD Standard deviation

Immunohistochemical (IHQ) Features

Ki-67 was positive in 19 (70.4%; [19/27] of the cases had this information) DGCT, and the PI ranged from < 1 to < 20%. All the 21 (100.0%) evaluated GCOC were Ki-67 positive, with the PI ranging from 2.9% to 61.8%. One benign case presented a Ki-67 PI of 40% in the basal/parabasal layers and 5% in stellate reticulum-like cells, but the authors did not showed the Ki-67 PI as a whole. Positivity for p53 was observed in 11 (84.6%; [11/13]) DGCT, ranging from focal to 68% labeling. Six (75.0%; [6/8]) GCOC showed a wide range of expression of p53.

CK14 (n = 13; 100.0%), AE1/AE3 (Pan-CK) (n = 10; 100.0%), CK19 (n = 18; 90.0%; [18/20]) were the most frequent cytokeratins (CKs) positivity labeling reported in neoplastic cells of DGCT, while AE1/AE3 (Pan-CK) (n = 5; 100.0%) and high molecular weight cytokeratin (HMC) (n = 4; 100.0%) were in GCOC. Ghost cells expressed CK19 (n = 1; 10.0%; [1/10]), CK6 (n = 6; 85.7%; [6/7]), and AE1/AE3 (n = 2; 66.7%; [2/3]) in DGCT and AE1/AE3 (Pan-CK) (n = 2; 66.7%; [2/3]) in GCOC.

β-catenin were positive in the neoplastic cells of all DGCT (n = 4; 100.0%) and GCOC (n = 2; 100.0%) examined. B-cell leukemia/lymphoma 2 (Bcl-2) antibody was positive in tumor cells of DGCT in 80.0% (n = 4; [4/5]) of the cases tested, and negative in the only (n = 1; 100.0%) GCOC tested. Ghost cells were always negative for Bcl-2. One GCOC was negative for Bcl-2 associated X (Bax) and positive for B-cell lymphoma-extra-large (Bcl-X_L) in the major component of neoplastic cells, while tumor cells adjacent to ghost cells were positive for Bcl-X_L and Bax, and ghost cells presented strong positivity for Bax and mild for Bcl-X_L.

Immunohistochemical features are detailed in Table 3.

Table 3.

Review of immunohistochemical data obtained from published cases and studies of DGCT and GCOC

References	n	Parenchymal cells					Ghost-cells	Dentinoid/osteodentin	Stromal cells
		PIa	p53	Cytokeratin	β-catenin	Other antibodies
DGCT
Urs et al. [64]  Urs et al. [86]	7	–	–	CK19 (+) [7C]; CK6 (+) [7C]	—	Amelogenin (+) [5C];	CK19 (+) [1C]; CK6 (+) [6C]; Amelogenin (+) [7C];	–	–
Inoue et al. [87]	1	Ki-67 (−)	(−)	CK19 (+); CK14 (+)	(+)	–	–	–	–
Rosa et al. [61]	6	Ki-67 (< 5%) [1C] Ki-67 (−) [5C] MCM-2 (< 5%) [5C] MCM-2 (−) [1C]	–	CK14 (+) [6C]; CK19 (+) [4C]	—	Amelogenin (+) [6C]; COL-1 (±)	CK14 (−) [6C]; CK19 (−) [6C]; Amelogenin (+) [6C]; DMP-1 (+) [4C]; COL-1 (+)	CK14 (−) [6C]; CK19 (−) [6C]; Amelogenin (−) [6C]; DMP-1 (−) [6C]; COL-1 (+) [6C]	COL-1 (+) [6C]
Soares et al. [88]	1	Ki-67 (< 1%)	–	AE1/AE3 (+); CK19 (+); CK14 (+); CK7 (+)	(+)	CD138 (+)	AE1/AE3 (−); CK7 (−); CK14 (−); CK19 (−)	–	CD34 (+); Vimentin (+)
Kanda et al. [89]	1	Ki-67 (9%)	–	–	–	–	–	–	–
Miwako et al. [55]	1	Ki-67 (4.1–4.4%)	(1.1–1.6%)	–	–	–	–	–	–
Walia et al. [90]	1	–	–	Pan-CK (+)	–	–	–	–	–
Jayasooriya et al.[4]	1	Ki-67 (< 1%)	–	MNF116 (+); CK14 (+); CK19 (< 5%)	–	–	–	–	–
Stojanov and Woo [35]	1	–	–	Pan-CK (+)	(+)	Vimentin (+)b	–	–	–
Soluk Tekkesin et al. [91]	1	–	–	CK1–3 (+)	–	–	–	–	–
Silva et al. [10]	1	Ki-67 (−)	–	–	–	Syndecan-1 (+)	–	–	Syndecan-1 (+)
Pulino et al. [92]	3	Ki-67 (< 20%) [3C] PCNA (< 20%) [1C] PCNA (≥ 50%) [2C]	(< 20%) [2C] (-) [1C]	–	–	–	–	–	–
Li et al. [93]	1	Ki-67 (13,5%)	–	CK5 (+); CK14 (+); CK18 (−)	–	–	–	–	–
Saghafi et al. [56]	6	PCNA (80–99%; ≈86.16%)	(40–68%; ≈54.3%)	–	–	–	–	–	–
Candido et al. [94]	1	–	–	Pan-CK (+); CK14 (+)	–	S100 (+)c; CD1a (+)d	–	–	–
Gong et al. [11]	7	Ki-67 (2.4–9.1%)	–	–	–	NF-κB p65 (+) [7C]; MMP-9 (+) [3C]	Ki-67 (−)	–	MMP-9 (+) [1C]
Iezzi et al. [65]	1	Ki-67 (+)	(±)	AE1/AE3 (+)	–	Bcl-2 (+)	AE1/AE3 (+); Ki-67 (−); Bcl-2 (−); p53 (−)	Ki-67 (−); Bcl-2 (−); p53 (−)	–
Yun et al. [95]	1	–	–	CK7 (+); CK19 (+)	–	–	–	–	–
Kim et al. [12]	1	–	–	AE1/AE3 (+)	(+)	Bcl-2 (−)	β-catenin (−), AE1/AE3 (+)	–	–
Fregnani et al. [67]	2	Ki-67 (< 1–1%; ≈0.5%) [1C] Ki-67 (−) [1C] PCNA (< 1–10%; ≈5%) [1C] PCNA (−) [1C]	–	AE1/AE3 (+) [2C]; CK8 (+) [2C]; CK14 (+) [2C]; CK19 (+) [2C]; 34βE12 (+) [2C]	–	Bcl-2 (+) [2C]	CK7 (−) [2C]; CK8 (−) [2C]; CK10 (−) [2C]; CK14 (−) [2C]; CK18 (−) [2C]; CK19 (−) [2C]; Bcl-2 (−) [2C]	–	–
Yoon et al. [96]	1	–	–	AE1/AE3 (+); CK19 (+)	–	S100 (−)	–	–	–
Mori et al. [97]	1	–	–	TK (+); KL1 (+); PKKL1 (±)	–	Involucrin (−); Involucrin (+)e; Vimentin (−)	–	Vimentin (−)	Vimentin (+)
Piattelli et al. [63]	1	Ki-67 (5–40%)	(±)	AE1/AE3 (+)	–	CD1a (+)d; Bcl-2 (+)	Ki-67 (−); Bcl−2 (−); p53 (−)	Ki-67 (−); Bcl-2 (−); p53 (−)	–
Lukinmaa et al. [62]	1	–	–	PKK1 (+); 34βE12 (+)	–	Tn-C (±)	PKK1 (−); 34βE12 (−); Tn-C (+)	Tn-C (±)	PKK1 (−); 34βE12 (−); Tn-C (+)
GCOC
Nel et al. [34]	1	Ki-67 (40%)	–	–	–	–	–	–	–
Araki et al. [98]	1	Ki-67 (35.7%)	(+)	CK14 (+); AE1 / AE3 (+)	(+)	–	–	–	–
Ohata et al. [15]	1	Ki-67 (15–32%; ≈20%)	(-)	AE1/AE3 (+); CAM5.2 (+)	(+)	α-SMA (−); S100 (−)	AE1/AE3 (+)	–	–
Remya et al. [27]	1	–	–	CK (+)	–	Vimentin (−); Desmin (−); SMA (−); CD34 (−)	–	–	–
Miwako et al. [55]	1	Ki-67 (2.9%); Ki-67 (22.8%)f	(36,2%)f	–	–	–	–	–	–
Park et al. [99]	1	Ki-67 (+)	–	Pan-CK (+)	–	–	–	–	–
Bose et al. [77]	1	Ki-67 (50–60%)	–	CK5 (±)	–	p63 (+)	–	–	–
Sukumaran et al. [83]	1	Ki-67 (50–60%)	–	CK (+)f; CK7 (−)f	–	p63 (+)f; TTF1 (−)f	–	–	–
Rappaport et al. [68]	1	Ki-67 (15–20%)	–	–	–	–	–	–	–
Ahmed et al. [84]	1	–	–	–	–	EGFR (+)	–	–	–
Del Corso et al. [6]	1	Ki-67 (≈10%)	(≈60%)	–	–	–	–	–	–
Silva et al. [10]	1	Ki-67 (49.3%)	–	–	–	Syndecan-1 (+)	–	–	Syndecan-1 (−)
Zhu et al. [54]	1	Ki-67 (61.8%)	–	–	–	MMP-9 (±)	Ki-67 (−); MMP-9 (±)	–	MMP-9 (+)
Li et al. [93]	1	Ki-67 (37.3%)	–	CK5 (+); CK14 (+); CK18 (−)	–	–	–	–	–
Kawai et al. [100]	1	–	–	CK19 (+)	–	Calretinin (+); CD56 (−)	–	–	–
Gong et al. [11]	5	Ki-67 (7.4–28%)	–	–	–	NF-κB p65 (+) [5C]; MMP-9 (+) [2C]	–	–	MMP-9 (+) [5C]
Motosugi et al. [58]	1	Ki-67 (4–28%; ≈16%)	(> 70%)	–	–	–	–	–	–
Sun et al. [7]	1	–	–	AE1/AE3 (+)	–	NSE (+); Vimentin (−); CEA (−); SMA (−); CD34 (−); S100 (−)	AE1/AE3 (+); NSE (+)	–	–
Zhang et al. [76]	5	–	–	–	–	SHH (+) [5C]; PTC (+) [5C]; SMO (+) [5C]; GLI 1 (+) [5C]	SHH (+) [5C]; PTC (+) [5C]; SMO (+) [5C]; GLI 1 (+) [5C]	–	–
Kim et al. [66]	1	–	–	AE1/AE3 (±)	–	Involucrin (±); Bcl-2 (−); Bcl-Xl (+); Bax (−); Bax (+)e	AE1/AE3 (−); Involucrin (−); Bcl-2 (−); Bcl-Xl (±); Bax (+)	–	–
Lu et al. [59]	3	Ki-67 (1–25%) [2C] Ki-67 (26–75%) [1C]	(1–25%) [1C] (26–75%) [1C] (76–100%) [1C]	LMC (+) [1C]; HMC (+) [3C];	–	CEA (−) [3C]; Vimentin (−) [3C]; S100 (−) [3C]; NSE (+) [3C]; Synaptophysin (−) [3C];	–	–	–
Folpe et al. [57]	1	PCNA (< 5%)	(−)	LMC (+); HMC (+)	–	Vimentin (+); CEAs (+); CEA (+); EMA (−); CD34 (−); Neurofilament (−); NGFR (−); S100 (+); CD57 (−); COL-IV (+)
Takata et al. [101]	4	PCNA (58.1–71.7%; ≈65,2%)	–	–	–	–	–	–	–

n Number of cases reported in the study; [nC] Number of cases in the study that obtained the referred labeling, (−) Negative; (+) Positive; (±) Weak/Focal/Rarely Positive; − Not available

^aProliferation Index

^bLabeling in tumor cells with plasmacytoid appearance

^cLabeling in Dendritic cells

^dLabeling in Langerhans cells

^eNucleated epithelial cells in ghost cell areas

^fLabeling in the metastatic site

Genomic Features

Single nucleotide alterations in the CTNNB1 gene were reported in codon 3 in one DGCT and in codon 33 in three GCOC. DGCT (n = 6) and GCOC (n = 2) lacked BRAFp.V600E mutation and presented a wild type BRAF gene. CREBBP and MLL2 gene mutations were also found in one GCOC.

Whole genome sequencing was performed in a single GCOC, showing triploidy, aneuploidies, structural and single nucleotide variations. Copy number alterations were reported in tumor suppressor genes (RB1, FHIT, PTEN, and RASSF4), ATM and CHEK2 genes, Sonic Hedgehog (SHH) (GLI1, SHH) and Notch (JAG1, DTX3, HEY1) signaling pathways members genes, and in oncogenes (AURKA, AKT1, GSK3B, and MYCN). Truncating mutation of the APC tumor suppressor gene, splice-site mutation of the UBR5 gene, and deletions in the TWIST1 gene were also noted. A novel fusion was found between the PTPRG-TCF4 genes in this tumor.

Genomic features are detailed in Table 4.

Table 4.

Review of genomic data obtained of published cases and research articles of DGCT and GCOC

References	n	Chromosomal alterations	Present gene mutations	Absent gene mutations
DGCT
Zhang et al. [72]	6	–	–	BRAF p.V600E
Ahn et al. [16]  Kim et al. [12]	1	–	CTNNB1 (ACT > TCT (Thr3Ser))	–
GCOC
Ohata et al. [15]	1	–	CTNNB1 (c.98C > G (p.Ser33Cys))	–
Bose et al. [77]*	1	FHL (RB1); FHD (FHIT); HTD (PTEN, RASSF4); CNA loss (ATM, CHEK2); CNA gain (JAG1, DTX3, HEY1, GLI1, SHH, AURKA, AKT1, GSK3B, MYCN); Fusion (PTPRG-TCF4)	APC (C > CT(V123Cfs)); UBR5 (C > T); TWIST1 (GGAT > G(I135del))	–
Rappaport et al. [68]	1	–	CTNNB1 (S33C); CREBBP (K1741*); MLL2 (S1997fs*44)	–
Diniz et al. [73]	2	–	–	BRAF p.V600E
Ahn et al. [16]	1	–	CTNNB1 (TCT > TAT(Ser33Tyr))	–

n Number of cases; FHL focal homozygous loss; FHD focal homozygous deletion, HTD heterozygous deletion, CNA copy number alterations

*Refer to the supplementary material of the referred article for the complete list of protein-coding single nucleotide variants and structural alterations detected in this case

Treatments and Outcomes

Definitive therapeutic modalities were grouped, and the surgical approaches were divided into conservative (simple excisions, curettages, and enucleations) or radical (partial and segmental resections). DGCTp were mainly approached conservatively (n = 28; 53.8%) and the DGCTc radically (n = 37; 48.7%). Therapeutic proposals for GCOC were mainly radical surgery (n = 22; 45.8%) or the combination of radical resection, neck dissection and/or (neo)adjuvant therapies (n = 18; 37.5%).

Higher recurrence rate was observed in DGCTc (n = 23; 30.3%) comparing with DGCTp (n = 1; 1.9%; Pearson chi-square test, p < 0,001; Φ = 0.420; OR 0.047; CI 0.006–0.368). An even higher percentage was noted in GCOC (n = 19; 39.6%), when compared with benign lesions (Pearson chi-square test, p-value = 0.043; Φ = 0.183; OR 0.450; CI 0.206–0.982). The average recurrence period was 50.73 months (± 66.71) in DGCT and 24.41 months (± 37.06) in GCOC (Mann–Whitney test, p value = 0.046, Effect size: 0.384). Metastases were noticed in 8 (16.7%) GCOC, and the lungs and/or pleura (n = 5; 10.4%) were the most affected sites. One (1.3%) maxillary DGCTc led the patient to death due to cranial invasion. Eight (16.7%) GCOC evolved to death due to local extension and/or metastasis. The 5-year overall survival rate for GCOC was 72.6% (Fig. 2).

Fig. 2.

Overall survival chart for GCOC

Therapeutic and prognostic features are detailed in Table 1.

Discussion

In our study, an increased incidence in males were found in both GCONs, affecting a widely variable age range, with an evident peak observed in DGCTp in elderly patients when compared to DGCTc and GCOC, as found in previous studies [5, 22, 26]. An Asian predilection has been reported in the literature for GCONs [3, 22, 27, 28]. This fact could not be confirmed in this review due to the lack of racial/ethnic data of the patients in the published cases. DGCTp and DGCTc were more frequently found in the mandible [3, 5, 22, 28]. The former was also reported once in the ethmoid sinus, and the authors suggested an origin from the respiratory mucosa since there was no underlying bone destruction [29]. In contrast, GCOC was twice as often reported in the maxilla, as previously mentioned [22].

DGCTp often appeared as mucosal firm painless nodular lesion measuring up to 3.0 cm, mimicking frequent reactive lesions of the oral mucosa [4, 30–32], being the main imaging finding, when present, a slight “cup-shaped” erosion of the underlying cortical bone [30, 32]. A great clinical resemblance was noted between DGCTc and GCOC in this review. Imaging findings proved to be a useful tool during its differentiation since malignant cases were more likely to present bone cortex destruction and ill-defined limits, while benign counterparts were mostly well-delimited and tend to cause cortical expansion. However, attention should be taken since some cases contrary to these presentations were observed [33, 34]. A wide range of benign and malignant odontogenic and non-odontogenic lesions should be considered in the differential diagnosis when mixed or radiolucent images of uni or multilocular appearance in the gnathic bones are observed [35, 36].

Microscopically, ghost cells (Fig. 3A) are also abundantly found in pilomatrixoma, craniopharyngioma, and could be seen as scattered cells in other odontogenic lesions, such as ameloblastoma, odontoma, and ameloblastic fibro-odontoma [14]. Furthermore, dentinoid material (Fig. 3B) has already been reported in the adenomatoid odontogenic tumor [9], complex odontoma, ameloblastic fibrodentinoma/fibro-odontoma,³¹ odontogenic carcinomas, and sarcomas [9, 37]. These previously mentioned lesions must be evaluated as potential differential diagnoses for GCONs since many of them can share clinical, radiographic, and histopathological characteristics. Interestingly, recurrent GCOC had a minor component of ghost cells in recurrences when compared to the previous lesion, with even the total absence of this component at the end of multiple events [38, 39].

Fig. 3.

Histopathological features found in ghost cell odontogenic neoplasms. Ghost cells (A). Dentinoid material (B). Ameloblastomatous epithelium (C). Basaloid cells (D). Adenoid-like/cribriform structures (E). Spindle cell differentiation (F)

According to WHO [8, 9], DGCT is composed of a predominant ameloblastomatous pattern (Fig. 3C) and a less prominent basaloid (Fig. 3D) epithelium, whereas GCOC is composed of a malignant hyperchromatic epithelium that varies from basaloid to large cells with vesicular nuclei, patterns predominantly proved in this review. However, besides this, other epithelial patterns were also observed, such as adenoid-like/cribriform structures (Fig. 3E), spindle (Fig. 3F), epidermoid and clear cell differentiation. Spindle, basaloid, and epidermoid differentiation patterns have also been reported in ameloblastoma and ameloblastic carcinoma, a spectrum of lesions with epithelium parenchymal similarities to GCONs [8, 9, 40–43]. Scattered or abundant clear cells were also reported in ameloblastomas and calcifying epithelial odontogenic tumors [41, 44]. Lesions with a mixture of ameloblastomatous epithelium with an adenoid-like component have been previously termed as adenoid ameloblastoma [45, 46], with some tumors presenting ghost cells and/or dentinoid material [45–49]. WHO stated that a proportion over 1–2% of ghost cells and dentinoid formation is useful for DGCT diagnosis [9]. To the best of our knowledge, this is the first paper that reviews a large sample of GCONs and describes the frequencies of the epithelial patterns, which includes the association of adenoid-like and ameloblastomatous areas.

DGCT and COC are known precursors lesions of GCOC [9]. Nevertheless, in this review, a previous diagnosis of ameloblastoma and calcifying epithelial odontogenic tumor, which were not recognized as precursor lesion, at the same site of GCOC, were observed [50, 51]. The diagnosis of borderline lesions during malignant transformation from COC and DGCT, is usually challenging. Sheik et al.[52] described a DGCTp with a possible focal malignant transformation in one area with mitosis and atypical cytological features. Ohata et al.[15] also showed overlapping of certain benign and malignant microscopic features in a predominantly solid lesion. Atypical areas were also previously associated with COC during its malignant transformation [53, 54]. These findings corroborate the need for serial assessment of histopathological sections during the diagnosis of GCONs due to the possibility of coexistence of benign and malignant components in the same lesion [8, 15, 27, 53].

The use of tumor PI and suppression markers to distinguish benign and malignant GCONs has already been assessed [10, 11, 15], but no cut-off points were stipulated to date [9]. Regarding the tumor PI, in our review, despite few exceptions [55–57], DGCT usually showed lower mean scores than GCOC, especially for the Ki-67 marker [10, 11]. Expression of p53 in the neoplastic cells had a wide range in GCOC, varying from less than 25% to more than 75% [58, 59] and was negative in two cases [15, 57]. In DGCT, p53 had lower values compared to GCOC. Nevertheless, one study observed values up to 68% in these lesions [56]. Overall, in our opinion, the lack of studies with a large sample and a unified methodological analysis of these scores brought the impossibility to obtain accurate cut-off values for GCONs.

Besides the importance to confirm the epithelial nature of GCONs, CKs expression can be important to understand the origin of these tumors, which remains unclear yet. CK7, CK14, and CK19 expression, as observed in some cases in our review, was previously reported in germinal dental tissues and other odontogenic tumors, reinforcing their odontogenic origin [4, 60, 61]. DGCTp was thought to arise from remnants of the dental lamina or alveolar mucosa surface epithelium [4, 32], and the continuity of the basal layer with the tumor parenchyma was reported in 9 (17.3%) cases. This feature was also present in cases with concomitant CK14, MNF116, and 34βE12 expression [4, 62]. The epithelial origin of ghost cells was confirmed through CKs expression [7, 12–15, 63–65].

The presence of ghost cells has been suggested to represent an aberrant form of terminal differentiation/keratinization of tumor cells, an event linked to the process of apoptosis [66]. IHC studies have already demonstrated the transition of anti-apoptotic to pro-apoptotic proteins in neoplastic to ghost cells [63, 65–67]. TUNEL assay technique assessed DNA fragmentation in epithelial cells adjacent ghost cells from both GCONs, corroborating this hypothesis [12, 66]. Involucrin expression in adjacent cells and negativity in ghost cells of a GCOC ratified the loss of normal keratinization and terminal differentiation in ghost cells [66].

The constitutive activation of the Wnt/β-catenin/TCF-Lef signaling pathway through the accumulation of β-catenin due to mutations in its coding gene CTNNB1 has been shown to have an essential association with the pathogenesis of ghost cell lesions, including COC, GCONs, pilomatrixoma, and craniopharyngioma, as well as other tumors [12, 13, 68]. Expression of β-catenin in neoplastic cells of GCONs with strong labeling specially located in nuclei of tumor cells adjacent to ghost cells [12], also suggests its participation in the development of ghost cells. The mutated CTNNB1 gene was thought to be a driver mutation in benign odontogenic ghost cell lesions and the main driver mutation with additional mutations for malignant progression in GCOC [12, 69]. Mutations in codons 3 and 33 were found in DGCT and GCOC [12, 15, 16, 68], respectively, and in codons 3, 4, 5, 32, 33, 37, 34 and 41 in COC lesions [16, 69–71]. Ameloblastomas with scattered ghost cells, in addition to BRAFp.V600G, also showed CTNNB1 gene mutation [70], whereas GCONs lack BRAF gene mutations [72, 73], reinforcing the importance of β-catenin gene mutation to ghost cells formation [70].

In vitro studies showed a key role of β-catenin in enhancing odontoblastic differentiation and dentin formation, presenting an explanation for the presence of dentinoid material in ghost cell odontogenic lesions [74]. Recently, some authors suggested an enameloid-like nature for this material due to its apparent secretion by peripheral ameloblast-like and mesenchymal cuboidal cells, expression of type 1 collagen (COL-1), and absence of amelogenin [61]. Plasmacytoid cells adjacent or interspersed with the dentinoid material were previously described by Stojanov and Woo[35] in a DGCT, which suggested a mesenchymal nature with secretory function of these cells due to positivity for vimentin and β-catenin and negativity for CKs [35].

Sonic Hedgehog (SHH) and Notch were cellular signaling pathways involved in the development of numerous neoplasms [75, 76]. SHH family members were highly expressed in GCOC [76], with genomic and transcriptomic investigations corroborating this finding [77]. Notch members had a genomic increased number of copies in GCOC [77], and were previously associated with ghost cell development [75]. Due to the limited number of cases studied to date, further investigations should be carried out to understand the alterations in these pathways in GCOC since they may represent potential therapeutic targets. Invasiveness of GCONs and metastatic capacity of GCOC have been demonstrated to be associated with the expression of matrix metalloproteinase 9 (MMP-9) in the neoplastic and stromal cells of GCONs [11, 54]. MMP-9 and Ki-67 were proposed to represent prognostic markers for GCOC, indicating the invasive, proliferative, and metastatic capacity [54].

Overall, due to the destructive behaviors and high recurrence rates found in DGCTc and GCOC, wide surgical resections with safe margins were recommended [78–80]. Prior treatment with local curettage and/or enucleation has already been associated with higher rates of recurrence in DGCTc [78]. Recurrence of DGCTp was extremely rare. Therefore, simple surgical excision with an additional underlying bone curettage seems to be an efficient approach [5]. GCOC may require adjuvant therapies depending on the medical team's evaluation [80], but studies have not been conducted to evaluate their indication and effectiveness. Regional lymph node and distant metastases were rare but were already documented in the brain, skull, pleura, lungs, and skin [50, 55, 68, 81–85]. A patient with recurrent DGCT developed the same lesion in the donor bone graft site, at ilium, collected for mandibular reconstruction [33]. The 5-year overall survival rate for GCOC was assessed only once, more than 20 years ago, and was found to be 73% [59], which is in line with our present finding of 72.6%. Therefore, a strict and long-term clinical and radiographic follow-up should be performed after treatment of all GCONs.

In conclusion, this is the largest systematic review with detailed diagnostic assessment criteria for GCONs and the first study to compare the histopathological, immunohistochemical, and genomic data of these published so far. The limitations of our study are mainly related to the lack of detailed information in publications, which may underestimate some of the results found. This reinforces the importance of publishing high-quality case reports and series. GCONs share numerous clinicopathological features and demand a careful histopathological evaluation taking into account the overlap with other odontogenic tumors and the capability of benign lesions to suffer malignant transformation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

GSV, DCC, RERM and KSGC conceptualized, wrote and revised the manuscript. GSV and PPM performed the extraction of the systematic review data. DCC and KSGC were the oral pathologists responsible for the support during the diagnostic confirmation of the cases. GSV and RERM conceptualized and performed the statistical analyzes of the systematic review data. All authors had given the final approval of the version to be published.

Funding

The authors G.S.V. and P.P.M. receive scholarships Grants from CAPES (Coordination for the Improvement of Higher Education Personnel).

Declarations

Conflict of interest

The authors declared no conflicts of interest.

Research Involving Human and Animal Rights

This article does not contain studies with human participants or animals performed by any of the authors.