Methylation of tumour suppressor genes in benign and malignant salivary gland tumours: a systematic review and meta-analysis

ABSTRACT

The aim of the present systematic review was to critically analyse the relationship between tumour suppressor genes (TSGs) promoter methylation, a potent mechanism of gene silencing, and the development of salivary gland tumours, as well as the possible effect on clinical/histological characteristics. Review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (registration ID CRD42020218511). A comprehensive search of Web of Science, Scopus, PubMed, and Cochrane Central Register of Controlled Trials was performed utilizing relevant key terms, supplemented by a search of grey literature. Newcastle-Ottawa Quality Assessment Scale (NOQAS) was used for the quality assessment of included studies. Sixteen cross-sectional and 12 case-control studies were included in the review, predominantly dealing with methylation in TSGs related to DNA repair, cell cycle, and cell growth regulation and differentiation. Quantitative synthesis could be performed on P16 (inhibitor of cyclin-dependent kinase 4a), RASSF1A (Ras association domain family 1 isoform A) and MGMT (O6-methylguanine DNA methyltransferase) genes only. It showed that P16 and RASSF1A genes were more frequently methylated in salivary gland tumours compared to controls (P = .0002 and P < .0001, respectively), while no significant difference was observed for MGMT. Additionally, P16 did not appear to be related to malignant transformation of pleomorphic adenomas (P = .330). In conclusion, TSG methylation is involved in salivary gland tumour pathogenesis and several genes might play a considerable role. Further studies are needed for a better understanding of complex epigenetic deregulation during salivary gland tumour development and progression.

Introduction

Salivary gland tumours (SGTs) represent a heterogeneous group of benign and malignant neoplasms exhibiting diverse yet overlapping and sometimes unpredictable clinical and histological characteristics; they also harbour multiple molecular changes [1,2]. Even though the vast majority of these neoplasms are benign (more than 80%), they have the ability to recur and/or transform into malignant lesions [1]. SGTs are rare entities, with malignant forms accounting for less than 0.5% of all cancers and around 6% of head and neck cancers [3]. In case of malignant SGTs the prognosis is extremely variable, with a high risk of cervical metastasis and soft tissue invasion and a 5-year overall survival rate of around 60% after surgery [4]. While the aetiology is still largely unknown, several risk factors have been suggested, including radiation, smoking, viral infections, immunosuppression, and genetic makeup [1,2].

Various mechanisms have been described in the pathogenesis of SGT: from chromosome translocations [5,6] and deletions [7], point mutations [8,9] and gene amplification [10,11] to epigenetic modifications [12,13], all leading to gene expression deregulation [14,15]. Epigenetic alterations of tumour suppressor genes (TSGs) are gaining more and more interest as possible contributors to salivary gland neoplastic transformation. They have been examined in a number of recent studies, but their role has not been fully elucidated yet. DNA methylation, an important epigenetic phenomenon that does not alter the sequence of nucleotide bases but modifies gene expression [16], was associated with different salivary gland tumours, both malignant and benign [17].

The inactivation of TSGs is crucial in the development of neoplasms, due to their involvement in cell cycle regulation and/or induction of apoptosis, DNA damage repair, and suppression of metastasis [18]. Previous studies have shown that aberrant promoter methylation is one of the most frequent TSG silencing mechanisms, a driving force taking place in the early phases of tumorigenesis [19]. Methylation of TSGs could contribute to salivary neoplasm progression alone or most likely in combination with other changes. Various cellular regulatory pathways were found to be affected by aberrant gene promoter methylation in different tumour types and subtypes [17]. It should be emphasized that DNA methylation is a reversible process [20] and as such, specific methylation patterns could represent not only diagnostic or prognostic markers but potential therapeutic targets as well.

The available scientific literature exposes inconsistent and conflicting results regarding the impact of TSG methylation in SGT. Therefore, the present systematic review aimed to critically assess the relationship of TSG promoter methylation and the development of salivary gland tumours, including its possible effects on clinical and histological features of these neoplasms.

Material and methods

To minimize the methodological bias and ensure transparency, precision, and integrity, this systematic review was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement guidelines [21]. The a priori defined review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (registration ID CRD42020218511).

Review question and eligibility criteria

The study was conducted to answer the following PICO question: Does tumour suppressor gene methylation play a role in the development of benign and malignant salivary gland tumours? To remove confounding factors, the review selected patients with benign or malignant salivary gland tumours but free of systemic diseases. Salivary gland tissues of healthy individuals or, alternatively, non-neoplastic salivary gland tissue from patients with a salivary gland tumour were used as controls. The primary outcome of concern was the presence or absence of tumour suppressor gene methylation in patients and, if applicable, in controls. The secondary outcomes of interest were possible associations of tumour suppressor gene methylation with sex, age, tumour location, size, histological type, stage, grade, neural invasion, presence of metastasis, or survival. Cohort studies, case-control studies, and cross-sectional studies, written in any language, were included. Papers presenting repeated results, reviews, meta-analyses, meeting abstracts, case reports, editorials, or letters were excluded. Studies that quantified tumour suppressor gene methylation, i.e., established the percentage of methylated CpG Islands within the promoter, without stating the number of methylated/unmethylated samples from SGT patients, in vitro studies, or those providing insufficient data for estimations were also excluded.

Search strategy

A comprehensive literature search was performed up to 16 March 2021, in Clarivate Analytics’ Web of Science, including Web of Science Core Collection (WoS), Korean Journal Database (KJD), Russian Science Citation Index (RSCI), and SciELO Citation Index (SciELO) [1980–2021], followed by Scopus [1960–2021], PubMed [1964–2021], and Cochrane Central Register of Controlled Trials (CENTRAL) [1996–2021], without language restrictions. A search strategy was jointly developed by the authorship team from a combination of relevant controlled vocabulary (Medical Subject Headings – MeSH, https://www.ncbi.nlm.nih.gov/mesh/) and the most common free keywords and synonyms for salivary gland tumours and gene methylation as the key concepts of interest. Preliminary searches of the specified databases were performed to produce and evaluate various information retrieval strategies, maximize sensitivity, and obtain the optimum structure. Database-specific syntax, Boolean operators (AND, OR), truncation, and proximity operators (NEAR, W) were applied. The final search strategy is detailed in Supplementary Table 1 for all examined sources. In parallel, to identify unpublished manuscripts, research reports, conference papers, doctoral dissertations, and other grey literature, complementary searches through OpenGrey, Google Scholar (first 100 returns), and other available digital repositories (e.g., Networked Digital Library of Theses and Dissertations, Open Access Theses and Dissertations, DART-Europe E-theses Portal – DEEP, Opening access to UK theses – EThOS) were completed. Finally, snowballing and screening of the reference lists of retrieved studies and relevant reviews were also performed to ensure the reliability of the data collected and inclusion of the relevant studies that may not have been identified through database and grey literature searches.

Study selection and data extraction

Studies identified in the literature search were imported into the Rayyan QCRI environment [22] for duplicate removal and further investigation. The study selection process for the systematic review was completed in two phases. Three independent reviewers (A.J., J.C., and J.J.) performed the initial screening of titles and abstracts from the search results to select studies that appeared to meet inclusion criteria. Papers that did not meet the eligibility criteria were excluded and full texts of initially selected articles were obtained for further assessment. In the second phase of study selection, three reviewers (A.J., J.C., and N.N.) independently evaluated full texts of studies identified as possibly being relevant in the initial screening stage. The obtained lists of relevant studies were compared, and disagreements regarding final study selection were resolved through discussion with a senior member of the review team (J.M.). For relevant studies with overlapping datasets or results, the most recent publication was included.

Data extraction was performed by two independent reviewers (J.C. and N.N.), uniformly completing a data collection form using Excel spreadsheets (Microsoft Excel, Redmond, WA, USA). Important information from each selected study was collected by one (J.C.) and cross-checked by a second reviewer (N.N.), confirming its accuracy. Data were extracted on the name and country of the first author, year of publication, analysed gene(s), type of study design, total number of participants, sample type, sample size per condition, tumour type, tumour location, molecular biology technique for methylation detection, and tumour suppressor gene methylation frequency by tumour type. Any disagreements were settled through discussion with a third reviewer (J.M.).

Quality assessment

Critical appraisal of the quality of included studies was assessed using the Newcastle-Ottawa Quality Assessment Scale (NOQAS) [23,24], concerning selection, comparability, and exposure biases, with a score of ≥7 denoting high quality. The assessment was performed by three independent reviewers (J.J., J.C., and A.J), and any disagreements were discussed and resolved by a fourth reviewer (N.N.). For any missing information, the corresponding authors of the individual studies were contacted.

To ensure the accuracy of the data analysis and establish the level of evidence of each outcome in this systematic review, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach was adopted [25].

Statistical analysis

All the relevant data were analysed qualitatively and quantitatively. A narrative synthesis of the main study findings is given. A quantitative meta-analysis was conducted using Review Manager Version 5.3. Based on the level of heterogeneity (inconsistency test), a random-effects model was used when heterogeneity was present, and a fixed-effect model if heterogeneity was not present. The forest plot was calculated considering odds ratios (ORs), 95% of confidence interval (CI) and p-values. Since less than 10 studies were included in each meta-analysis, the assessment of the publication bias via funnel plot was not advised [26]. The geo-mapping of the included studies by country was completed using the R package rworldmap version 1.3.6 [27].

Results

Study selection

Initial database search, followed by the removal of 167 duplicates, yielded a total of 193 studies for potential inclusion in the present systematic review. After the first screening of titles and abstracts, 160 articles did not satisfy inclusion criteria and were therefore excluded. Following a comprehensive full-text examination of 34 remaining studies, six studies were excluded due to reasons detailed in Supplementary Table 2. Finally, 28 articles were included in the present qualitative analysis. The PRISMA flow diagram for the selection process based on the presented exclusion criteria is shown in Figure 1.

Figure 1.

PRISMA Flow diagram describing the study selection process. n – number of hints.

The principal features of the included studies are presented in Table 1. Out of 28 studies included in the qualitative synthesis, 16 were conducted under a cross-sectional study design and 12 were case-control studies. The majority of studies originated from China and Japan (seven studies each), three studies were from the United States, while Brazil, Germany, Italy, and Serbia were represented by two and Taiwan, Turkey, and the United Kingdom by one study each (Supplementary Figure 1). The majority of studies used methylation-specific polymerase-chain reaction (MS-PCR) for the analysis of TSGs methylation status, while the most commonly used tissue samples were formalin-fixed paraffin-embedded (FFPE) (Table 1). Both major and minor salivary glands were represented in the included studies.

Table 1.

Basic characteristics of eligible studies.

First author’s name	Year of publication	Country	Gene	Study type	Sample type	Tumour location	Number of participants	Methylation detection method
Augello, C.[38]	2006	Italy	p16	cross-sectional	FFPE	N/A	33 SGTs	MS-PCR
Daa, T.[68]	2008	Japan	p15, p18, p19, p21, p27.	cross-sectional	FFPE	N/A	34 SGTs	MS-PCR
Gomes, C.C.[69]	2011	Brazil	WWOX exon1	cross-sectional	FFS	N/A	27 SGTs	MS-PCR
Hu, Y.H.[39]	2011	China	p16	cross-sectional	FFPE	42 major, 8 minor	50 SGTs	MS-PCR
Kishi, M.[28]	2005	Japan	RB1, p21, p27, PTEN, MGMT	cross-sectional	FFPE	24 major, 12 minor	36 SGTs	MS-PCR
Lee, E.S.[29]	2008	USA	RARβ2, RASSF1A, MGMT	cross-sectional	FFS	major, minor	69 SGTs	MS-PCR, PS
Li, J.[30]	2005	China	p16, RASSF1A, DAPK, MGMT	cross-sectional	FFPE	27 major, 33 minor	60 SGTs	MS-PCR
Mariano, F. V.[35]	2016	Brazil	RASSF1A, PTEN	cross-sectional	FFPE	11 major, 2 minor	15 SGTs	MLPA
Maruya, S. I.[40]	2003	Japan	p16	cross-sectional	FFPE	8 major, 14 minor	22 SGTs + 3 R&M	MS-PCR
Maruya, S. I.[45]	2004	Japan	CDH1	cross-sectional	FFPE	9 major, 14 minor	23 SGTs + 3 R&M	MS-PCR
Nikolic, N.[12]	2015	Serbia	p14, p16	cross-sectional	FFPE	parotid	60 SGTs	MS-PCR
Nishimine, M.[41]	2003	Japan	p14, p16	cross-sectional	FFPE	N/A	36 SGTs	MS-PCR
Schache, A.G.[36]	2010	United Kingdom	p16, CYGB, RASSF1A, RARβ2, WT1, TMEFF2	cross-sectional	FFPE	N/A	59 SGTs	MS-PCR (qMS-PCR)
Toso, A.[31]	2009	Italy	MGMT, p73, DAPK	cross-sectional	TT	6 major, 2 minor	8 SGTs	MS-PCR
Xia, L.[47]	2018	China	CDH1	cross-sectional	FFPE	29 major, 8 minor	37 SGTs	BS
Zhang, C.Y.[46]	2007	China	CDH1	cross-sectional	FFPE	27 major, 33 minor	60 SGTs	MS-PCR
Durr, M.L.[32]	2010	USA	p16, stratifin 14-3-3s, RASSF1A, RARβ2, DAPK, MGMT, AIM1, APC, b-catenin, DCC, FHIT, GSTP1, HIC1, Mint1, MLH1, PGP9.5, THBS1, TIMP3, TMS1	case-control	FFPE	59 major, 19 minor	78 SGTs/17 NSGs	qMS-PCR
Ge, M. H.[49]	2011	China	RUNX3	case-control	91 FFPE, 23 FFS	44 major, 70 minor	114 SGTs/114 NSGs	qMS-PCR
Guo, X.L.[42]	2007	China	p16	case-control	FFS	SGT: 21 major, 17 minor; NSG – 4 major, 2 minor	38 SGTs/6 NSGs	MS-PCR
Lee, C. H.[51]	2010	Taiwan	APC, SFRPs	case-control	FFPE	N/A	46 SGTs/25 NSGs	MS-PCR
Nikolic, N.[13]	2018	Serbia	p14, p16, TP53	case-control	FFPE	28 major, 7 minor	35 SGTs/10 NSGs	MS-PCR
Ozen, F.[48]	2020	Turkey	CDH1	case-control	FFPE	parotid	23 SGTs/10 NSGs	MS-PCR
Sasahira, T.[50]	2011	Japan	RUNX3	case-control	FFPE	N/A	24 SGTs	MS-PCR, BS
Scesnaite, A.[33]	2014	Germany	MGMT	case-control	FFPE	major, minor	36 SGTs/19 NSGs	PS
Suzuki, H.[43]	1998	Japan	p16	case-control	FFPE	parotid	4 SGTs/4 NSGs	MS-PCR
Weber, A.[44]	2002	Germany	p14, p16	case-control	FFS	parotid	42 SGTs/42 NSGs	MS-PCR, RE-PCR
Williams, M.D.[34]	2006	USA	DAPK, MGMT, RARβ2, RASSF1A	case-control	FFS	malignant: 67 major, 12 minor	102 SGTs/29 NSGs	MS-PCR
Zhang, C.Y.[37]	2014	China	RASSF1A	case-control	FFPE	53 major, 114 minor	167 SGTs/50 NSGs	BS, MS-PCR

WWOX – WW domain-containing oxidoreductase; RB1 – retinoblastoma protein; PTEN – Phosphatase and tensin homolog; MGMT – O6-methylguanine-DNA-methyltransferase; RARβ2 – retinoic acid receptor β2; RASSF1A – RAS-associated domain family protein 1A; DAPK – death-associated protein kinase; CDH1 – Cadherin-1; CYGB – cytoglobin; WT1 – Wilms’ tumour 1; TMEFF2 – Transmembrane Protein with EGF Like and Two Follistatin Like Domains 2; AIM1 – absent in melanoma-1; DCC – deleted in colorectal carcinoma; FHIT – fragile histidine triad; GSTP1 – glutathione S-transferase P1; HIC1 – hypermethylated in cancer-1; Mint1 – methylated in tumour-1; MLH1 – mismatch repair protein; PGP9.5 – protein gene product 9.5; THBS1 – thrombospondin-1; TIMP3 – tissue inhibitor of metalloproteinase-3; TMS1 – target of methylation induced silencing-1; RUNX3 – Runt-related transcription factor 3; APC – Adenomatous polyposis coli; SFRP – Secreted frizzled-related protein; FFPE – formalin fixed paraffin embedded; FFS – freshly frozen samples; TT – tumour tissue; SGT – salivary gland tumour; NSG – normal salivary gland; BS – bisulphite sequencing; PS – pyrosequencing; R&M – recurrent and metastatic.

Main outcomes

The most represented malignant tumour types in the studies included in this systematic review were adenoid cystic carcinoma (ACC), mucoepidermoid carcinoma (MEC), carcinoma ex pleomorphic adenoma (CexPA), and salivary duct carcinoma (SDC), while the benign tumours were pleomorphic adenomas (PA) and Warthin tumours (WT) (Table 2).

Table 2.

Methylation frequencies of the tumour suppressor genes analysed in at least two studies.

	analysed gene(s)	study	sample size	TUMOUR (Nmeth/Ntot)
				malignant													benign
				ACC	MEC	CexPA	SDC	Acinic CC	BCAC	SCC	ADC	ME	AC	MALToma	MPA	SGC*	PA	WT
Cell cycle regulation	p16	Augello (2006)	33	4/4		1/1							0/2				4/28
		Durr (2010)	78	0/17	2/17		6/18						0/2				1/26
		Guo (2007)	38		13/38
		Hu (2011)	100			18/50											4/50
		Li (2005)	60	28/60
		Maruya (2003)	22	4/22
		Nikolic (2015)	60			6/10											35/50
		Nikolic (2018)	35		21/35
		Nishimine (2003)	36	3/20	0/7	1/1	0/1	0/1	0/1	0/2					0/1
		Schache (2010)	59			4/31											0/28
		Suzuki (1998)	8			0/4											0/4
		Weber (2002)	42														12/42
	p21	Daa (2008)	34	24/26
		Kishi (2005)	36	0/17	0/7	0/2	0/1	0/3	0/1	0/3	0/2
	p27	Daa (2008)	34	9/34
		Kishi (2005)	36	2/17	1/7	0/2	0/1	0/3	0/1	0/3	0/2
	p14	Nikolic (2015)	60			9/10											36/50
		Nikolic (2018)	35		35/35
		Nishimine (2003)	36	5/20	0/7	0/1	0/1	0/1	0/1	0/2					1/1
		Weber (2002)	42														1/42
	DAPK	Durr (2010)	78	6/17	4/17		12/18										3/26
		Li (2005)	60	16/60
		Toso (2009)	8											5/8
		Williams (2006)	102	2/26	0/18		2/21	0/14				0/2					0/12	4/9
	RASSF1A	Durr (2010)	78	6/17	1/17		12/18										14/26
		Lee (2008)	69	4/25	1/17	2/6	2/8	6/13
		Li (2005)	60	25/60
		Mariano (2016)	15			2/5											4/10
		Schache (2010)	59			16/31											2/28
		Williams (2006)	102	4/26	1/18		10/21	6/14				0/2					2/12	0/9
		Zhang (2014)	167	59/167
	PTEN	Kishi (2005)	36	0/17	0/7	0/2	0/1	0/3	0/1	0/3	0/2
		Mariano (2016)	15			1/5											0/10
Cell growth and differentiation	RARβ2	Durr (2010)	78	3/17	4/17		14/18										2/26
		Lee (2008)	69	0/25	3/17	1/6	2/7	0/13
		Schache (2010)	59			0/31											0/28
		Williams (2006)	102	1/26	3/18		6/21	0/14				0/2					0/12	0/9
DNA repair	MGMT	Durr (2010)	78	0/17	0/17		2/18										0/26
		Kishi (2005)	36	0/17	2/7	0/2	0/1	0/3	0/1	0/3	0/2
		Lee (2008)	69	0/25	2/17	0/6	0/8	0/13
		Li (2005)	60	4/60
		Scesnaite (2013)	36													10/36
		Toso (2009)	8											0/8
		Williams (2006)	102	1/26	3/17		0/20	0/14				0/1					1/12	3/9
Cell-cell adhesion	CDH1	Maruya (2004)	23	16/23
		Ozen (2020)	23														6/23
		Xia (2018)	37			21/37
		Zhang (2007)	60	34/60
	APC	Durr (2010)	78	6/17	2/17		15/18										9/26
		Lee (2010)	46		16/46
Transcription
		Ge (2011)	114	58/114
	RUNX3	Sasahira (2010)	24	6/8	7/8												2/8

DAPK – death-associated protein kinase; RASSF1A – RAS-associated domain family protein 1A; PTEN – Phosphatase and tensin homolog; RARβ2 – retinoic acid receptor β2; MGMT – O [6]-methylguanine-DNA-methyltransferase; CDH1 – Cadherin-1; APC – Adenomatous polyposis coli; RUNX3 – Runt-related transcription factor 3; Nmeth – number of methylated samples; Ntot – total number of samples.

* – histological type not specified.

The majority of studies have investigated genes related to DNA repair, regulation of cell cycle and cell growth and differentiation. MGMT gene was analysed for the presence of methylation in seven studies [28–34]. In most of them, a low frequency of MGMT methylation was reported (from 0% to 8%). However, Scesnaite et al. found MGMT methylation in 28% of SGC cases [33]. RASSF1A methylation was also analysed in seven studies [29,30,32,34–37] and appeared to be a more frequent event in both malignant and benign salivary gland tumours than MGMT methylation (from 22% to 42%). P16 was the most analysed TSG [12,13,30,32,36,38–44], and the reported incidences of its methylation varied greatly: in malignant tumours from 0% [43] to 100% [13] (Table 2) and in benign from 0% [36,43] to 70% [12]. The frequency of DAPK methylation also varied considerably between the studies, from 9% [34] to 63% [31]) but was , generally, rather low (Table 2).

RARβ2 gene was analysed in four studies, and the highest frequency of methylation was reported in SDC (78%) in the study of Durr et al. [32]. However, Schache et al. did not find any methylated sample among their examined cases (CexPA or PA) [36].

Another common subject of investigation was cell–cell adhesion and signalling in epithelial tissues, involving CDH1 and APC genes. CDH1 gene methylation was more frequent in malignant (in ACC – 70% [45] and 57% [46]; CexPA – 57% [47]) than in benign salivary gland tumours (26% [48]). The frequency of APC methylation also varied considerably, from 12% in MEC [32] to 83% in SDC [32].

Methylation of the transcription regulator, encoded by RUNX3 gene, seemingly plays an important role in the pathogenesis of ACC and MEC as judged by its high frequency: 50.9% to 75% in ACC [49,50] and 87.5% in MEC [50].

Details regarding the genes that were explored in only one study each are given in Supplementary Table 3.

Secondary outcomes

Seventeen studies [12,13,28–30,33,34,37,39,40,42,45–49,51] assessed the possible association with sex, age, tumour location, size, histological type, stage, grade, neural invasion, presence of metastasis, or survival. The majority of studies did not report significant associations or correlations (Figure 2). CDH1 methylation was associated with female sex in one study [47], while two other studies reported no association with the sex [46,48]. It was also correlated with higher histological grade in two studies [46,47], while one reports a lack of correlation [45]. Opposing findings were also reported for the clinical stage [46,47] and survival [45,47]. CDH1 methylation was [46]/was not [47] associated with perineural invasion. Controversial results were also reported for RASSF1A methylation in relation to the clinical characteristics [29,30,37]. While Li et al. [30] and Zhang et al. [37] found a correlation with the clinical stage, it was not reported by Lee et al. [29]; opposing findings were also given regarding the correlation with histological grade and the presence of metastasis [29,30].

Figure 2.

A diagram depicting association of gene methylation frequency with patients’ sex, age, tumour location, tumour size, stage, grade, the presence of metastasis or neural invasion, histological type or survival. The number of boxes per bar corresponds to the number of studies assessing the association; green (YES) indicates the presence of association, red (NO) the absence of association and the blank box (NE) implies the association was not evaluated in the study. APC – Adenomatous polyposis coli; CDH1 – Cadherin-1; DAPK – death-associated protein kinase; MGMT – O6-methylguanine-DNA-methyltransferase; RARβ2 – retinoic acid receptor β2; PTEN – Phosphatase and tensin homolog; RASSF1A – RAS-associated domain family protein 1A; RB1 – retinoblastoma protein; RUNX3 – Runt-related transcription factor 3; SFRP – Secreted frizzled-related protein.

Meta-analyses of methylation frequencies: salivary gland tumours versus normal salivary glands

Pooled data from Durr et al., Guo et al., Nikolic et al., Suzuki et al., and Weber et al. [13,32,42–44] were used to perform the quantitative synthesis in order to evaluate the potential association of p16 promoter methylation with the development of salivary gland tumours. Figure 3(a) shows the forest plot for the distribution of methylation-positive samples between the salivary gland tumour (SGT) and normal salivary gland (NSG) groups. Based on the available data from these five studies, significant increase of p16 methylation frequency was observed in SGT (OR = 9.43; 95% CI 2.90–30.62; P = 0.0002; I² = 0%, Figure 3(a)).

Figure 3.

Forest plots of the (a) p16, (b) RASSF1A and (c) MGMT gene promoter methylation frequency distribution between salivary gland tumours (SGTs) and normal salivary glands (NSGs).

Similar quantitative synthesis was performed using the pooled data from Durr et al., Williams et al., and Zhang et al. [32,34,37] to evaluate the association of RASSF1A methylation with the SGT development (Figure 3(b)). Significantly higher frequency of RASSF1A methylation was observed in SGTs compared to NSGs (OR = 31.98; 95% CI 6.30–162.39; P < 0.0001; I² = 0%, Figure 3(b)).

In contrast, based on the available data from three studies [32–34], no significant difference was found between the groups regarding the MGMT methylation (P = 0.140), also with 0% heterogeneity between the studies (Figure 3(c)).

Meta-analysis of methylation frequency: PA versus CexPA

Another quantitative synthesis was performed using the pooled data from Augello et al., Nikolic et al., and Schache et al. [12,36,38], with the aim of assessing the association of p16 promoter methylation with the malignant transformation of PA into CexPA. Figure 4 illustrates the forest plots for the distribution of p16 methylation in PA and CexPA samples. It should be emphasized that the studies analysing malignant and benign components of the same CexPA tissue were excluded from the quantitative synthesis [39,43]. No statistically significant difference was observed between the groups (P = 0.330) and the heterogeneity between the studies was notable (I [2] = 59%), although nonsignificant (P = 0.090). However, imprecision remained substantial with wide CI(s) for both analyses.

Figure 4.

Forest plot of the p16 gene methylation frequency distribution between carcinoma ex pleomorphic adenoma (CexPA) and pleomorphic adenoma (PA) patients.

Subgroup analysis

No subgroup analysis was required in any of the quantitative analyses since the heterogeneity was not significant.

Quality assessment

The detailed results of the evaluation of the quality of the 16 cross-sectional and 12 case-control studies included in this review are presented in Supplementary Tables 4 and 5, respectively.

Based on the NOS scale, the overall methodological quality is high for all cross-sectional studies (Supplementary Table 4), while eight case-control studies scored 7 stars or more and were ranked as high quality (Supplementary Table 5). Three of the remaining studies were moderate and one was low quality. Deficiencies identified in the cross-sectional studies were mainly related to unjustified sample size or to the used statistical test that was not completely or appropriately described, while the main deficiencies in case-control studies were related to the inappropriate selection of controls or failure to describe the recruitment process.

There was a low certainty of evidence per GRADE (Supplementary Tables 6 and 7). This indicates that it is very likely that further research might change the estimate of the effect.

Discussion

Tumour suppressor genes are involved in various biological processes, but their most important cancer-preventing functions include control of cell cycle, cell growth and differentiation, DNA repair, cell-to-cell adhesion, and transcription [18]. Inactivation of these genes grants the cells viability and growth advantage, thus contributing to neoplastic transformation and tumour development [52]. Our review showed that numerous TSGs have aberrant promoter methylation patterns in SGTs. Certain genes, however, seem to be affected more often than others. Genes that were investigated in more than one study comprise only a fraction of what is known to be a cancer-related gene alteration network, involved in neoplastic transformation in the vast majority of human cancers (Figure 5).

Figure 5.

A broad network of linkages between most frequently analysed genes affected by gene methylation in salivary gland tumours; drawn by GeneMANIA (Warde-Farley et al. 2010). MGMT – O-6-methylguanine-DNA methyltransferase; CDKN – cyclin dependent kinase inhibitor; RUNX – runt related transcription factor; RARB – retinoic acid receptor beta (RARβ2); PTEN – phosphatase and tensin homolog; CDH – cadherin; DAPK1 – death associated protein kinase 1; RASSF – Ras association domain family member; APC – adenomatous polyposis coli; E4F1 – E4F transcription factor 1; RNF114 – ring finger protein 114; KCTD1 – potassium channel tetramerization domain containing 1; TP63 – tumour protein p63; TPTE – transmembrane phosphoinositide 3-phosphatase with tensin homology; DNAJC6 – DnaJ heat shock protein family (Hsp40) member C6; DSC – desmocollin.

P16 gene, encoding the p16INK4a protein, regulates the cell cycle via p53 and RB1 pathways and is the second most commonly altered TSG in all cancers. Apart from homozygous deletions, promoter methylation is the most frequent alteration of p16, typically resulting in loss of function [53]. This review identified p16 as the most analysed TSG in SGTs (Table 2), with extremely diverse methylation frequencies in both benign (from 0% to 70%) [12,32,36,38,39,43,44] and malignant (from 0% to 100%) tumours [12,13,30,32,36,38–43]. No association was found between p16 methylation and clinicopathological features of SGTs in any of the studies (Figure 2). The meta-analysis of the association of p16 and salivary gland tumour development demonstrated significantly increased methylation frequency in SGTs compared to normal salivary glands (NSGs) (OR = 9.43; 95% CI 2.90–30.62; P = 0.0002; I² = 0%, Figure 3(a)). However, the second quantitative synthesis in the present study failed to establish a link between p16 methylation and malignant transformation of PA into CexPA (Figure 4). Previous meta-analyses have associated higher frequencies of p16 methylation with different head and neck cancers, including OSCC, and proposed this event as a possible diagnostic and prognostic marker [54–57].

DAPK1 gene, encoding Ca+/calmodulin-regulated serine/threonine kinase, is the most extensively studied gene of the death-associated protein kinase (DAPK) family. The list of its functions includes the regulation of apoptosis, autophagy, and inhibition of metastasis [58]. Promoter methylation of DAPK was previously associated with decreased expression of DAPK and was found in several types of cancer [57,59]. Our analysis demonstrated that DAPK methylation was not associated with demographic or clinicopathological features of SGTs (Figure 2) and its frequency was below 50%, except in the study of Toso and co-workers [31].

RASSF1A is involved in the stimulation of mitotic arrest, DNA repair, apoptosis, as well as in the control of cell cycle and cell migration. The frequency of RASSF1A promoter methylation was increased in most cancer types and inversely correlated with its expression [60]. The impact of RASSF1A methylation on the clinical characteristics of SGTs appears debatable [29,30,34,37]; nonetheless, it is a very frequent DNA modification both in benign and malignant tumours [29,30,32,34–37] pointing to its importance in SGT pathogenesis. This view was additionally corroborated by the finding of much higher methylation frequency of RASSF1A in SGTs compared to NSGs, shown in the quantitative synthesis (OR = 31.98; 95% CI 6.30–162.39; P < 0.0001; I² = 0%, Figure 3(b)).

The retinoic acid receptor-beta 2 (RARβ2) gene is involved in a range of cancer-suppressing effects of retinoids, including control of differentiation and cell growth. Frequencies RARβ2 promoter methylation in SGTs ranged from 0% to 78% [29,32,34,36] and were, in some instances, associated with tumour grade, histological type, and patients’ survival [34].

The MGMT gene encodes a DNA repair protein essential for the prevention of malignant transformation. However, MGMT silencing by methylation does not appear to be excessively important in SGTs development. The frequency of MGMT methylation was mostly low – from 0% to 8% [28–32,34], with the exception of the study by Scesnaite et al., with 28% of methylated SGC samples [33]. No correlation between MGMT methylation and clinicopathological tumour features could be established (Figure 2) [28–30,33]. Also, no difference was found between SGTs and NSGs in our meta-analyses (Figure 3(c)).

CDH1 gene encodes a transmembrane protein (E-cadherin) and exerts its tumour-suppressing role through connecting epithelial cells, which is of particular importance in SGTs with an epithelial component. Although E-cadherin loss, arising mostly as the consequence of epigenetic modifications, may not be the sole cause of epithelial–mesenchymal transition [61], it has frequently been associated with poor prognosis and survival in patients with various cancers [62]. In salivary gland tumours higher frequencies (varying between 57% and 70%) of CDH1 methylation were found in malignant lesions [45–47], compared to benign (only 26%) [48]. Regarding the association of methylation with demographic and clinicopathological characteristics, such as sex, histological grade, clinical stage, and perineural invasion, reported findings are very conflicting [45–48].

APC (Adenomatous polyposis coli) gene is an important WNT-signalling regulator involved in apoptosis and cell cycle arrest. Methylation of APC promoter was observed in different tumours and shown to inhibit gene expression [63]. Only two studies dealing with its methylation frequency in SGTs were included in this review. In mucoepidermoid carcinoma 35% of tumours were methylated [51]. In the study of Durr et al., which included different SGTs, the frequency of methylation varied from 12% to 83% depending on the tumour type, but without any association with clinicopathological characteristics [32].

RUNX3 gene is one of the three mammalian Runt-domain transcription factors (TFs), mostly linked to solid tumours. Its role, however, remains somewhat controversial since it exerts both pro- and anti-tumorigenic effects, even in the same cancer types [64,65]. RUNX3 methylation in SGTs appears to exhibit a pro-tumorigenic role as higher frequencies of methylation were present in malignant [49,50] than in benign SGTs [50]. Tumour stage, neural invasion, metastasis, and patient survival were associated with RUNX3 methylation (Figure 2).

TSG methylation frequencies, their association with clinicopathological tumour features and their respective contribution to the risk of SGTs development displayed a great diversity, which could be, at least in part, attributed to differences between studies, including the study design itself (cross-sectional and case-control studies), participants’ age, sex, ethnicity and type of salivary gland tumours. Also, differences in sample size and methylation detection approaches could lead to somewhat discrepant results. Methodological quality assessment (NOS scale) identified certain deficiencies in the study designs such as unjustified and therefore possibly inadequate sample sizes, inappropriate control sampling protocols, and unexplained statistical tests used to analyse the data. All these inadequacies should be taken into consideration in future investigations, in order to gain a clearer perspective, especially regarding promoter methylation of the TSGs emphasized in the present systematic review as putative biomarkers of SGT etiopathogenesis. However, despite all these issues encountered in the included studies, our review has identified several TSGs whose aberrant promoter methylation may contribute to SGTs development and progression. This finding should be corroborated in future, carefully designed primary studies, on more adequate and justified sample sizes, focusing on specific tumour subtypes and addressing the tumour clinicopathological features.

Some limitations of the meta-analyses must be stressed. No more than five studies with relatively small sample sizes could be included in each quantitative synthesis. This has resulted in high imprecision, as demonstrated in Figures 3 and 4 with wide CI; hence further research could have an impact on the estimated effect. Moreover, although the included studies spanned across the Asian, South and North American, and European continents, the ethnicities of the subjects were mostly unknown. Literature evidence suggests that methylation frequencies are related to ethnic background [66,67], pointing to the necessity of taking this fact into account in future studies.

Conclusion

The present systematic review and meta-analysis suggest that tumour suppressor gene methylation represents an important event in the pathogenesis of salivary gland tumours. Additional well-designed, multicentric, and multiethnic studies with a larger sample size are needed for a better understanding of complex epigenetic deregulation during salivary gland tumour development and progression.

Funding Statement

This work was supported by the Ministry of Education, Science and Technological Development of Republic of Serbia [451-03-9/2021-14/200129]

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors hereby confirm that the data supporting the findings of this systematic review and meta-analysis are available within the article [and/or] its supplementary materials.

Supplementary material

Supplemental data for this article can be accessed here.