Epigenetic alterations in head and neck cancer: a brief update

Abstract

DNA methylation of tumour suppressor genes is the most well-studied epigenetic alterations in head and neck cancer. The tumour suppressor genes CDKN2A, RASSF1, and TIMP3 are the most frequently investigated, but the methylation status has been analysed in more than another dozen genes, for example MGMT. In oral squamous cell carcinoma (OSCC) methylation of MGMT, DAPK, and CDKN2A are promising biomarkers of prognostic value. Inhibition of LSD1, encoding a histone demethylase, attenuates the development and growth of OSCC. Methylation of TIMP3 in sinonasal adenocarcinoma (intestinal type) is associated with a significant worse survival, an association not seen in sinonasal squamous cell carcinoma. Olfactory neuroblastoma can be distinguished into four unique subgroups by methylation profiling. Methylation of RASSF1 is seen in NUT carcinoma, and significantly higher RASSF1 methylation is found in SMARCB1/INI1-deficient tumours compared to the less aggressive SMARCB1/INI-proficient tumours. Genome-wide methylation profiling in combination with IDH2 mutation status suggests that tumours with undifferentiable SNUC morphology can be classified into for subgroups. Most salivary gland carcinoma subtypes have specific epigenetic signatures. Four of the most common subtypes, adenoid cystic carcinoma (ADCC), mucoepidermoid carcinoma (MEC), acinic cell carcinoma (ACC), and carcinoma ex pleomorphic adenoma (CXPA) have all methylation of RASSF1A, two (MEC and ADCC) also of TIMP3 and two (MEC and CXPA) of p16INK^4a. A methylation landscape of 20 salivary gland tumours (SGTs) is nowadays available.

Introduction

During the last decades a wealth of studies on genetic aspects of tumour development and progression have led to a tremendous increase in knowledge, particularly regarding mutations, chromosome translocations, and pathways that activate oncogenes and inactivate tumour suppressor and DNA repair genes. Besides genetic changes, epigenetic alterations have increasingly been highlighted as crucial in the initiation and progression of human tumours [1]. Epigenetic alterations refer to stable and heritable changes in gene expression without changes in the DNA sequence [2] and denote the way that genes interact with the environment, in order to generate individual phenotypes [3]. Epigenetic modifications include DNA methylation, covalent histone modifications, chromatin remodelling, and effects of non-coding RNAs and polycomb proteins on gene expression [4–6]. DNA methylation is the principal epigenetic factor and is mediated by DNA methyltransferases involving the methylation of cytosine in CpG dinucleotides. CpG islands are regions of DNA found near the start sites of genes (5’ promotor region) and are rich in the dinucleotide sequence "CpG" (cytosine followed by guanine). They have a high guanine-cytosine content compared to the rest of the genome. Hypermethylation of the CpG islands on gene promoters stimulates carcinogenesis, for example via silencing of tumour suppressor genes (TSG) like e.g. CDKN2A (p16^INK4a/p14^ARF), and RB1 [7]. Aberrant DNA methylation represents an epigenetic hallmark strongly associated with tumour initiation and progression leading to inactivation or reduced expression of TSGs and activation of oncogenes [8]. Silencing of TSGs, commonly through epigenetic mechanisms like promoter methylation, is a critical early event in many cancers. The methylation status of the hypermethylated oncological region (THOR) of human telomerase reverse transcriptase (TERT) has been proven to be a robust diagnostic and prognostic biomarker in a number of human tumours, for example in cancers of the prostate, pancreas and breast, as well as of meningiomas [9–14]. Contrary to gene mutations, epigenetic events are reversible and therefore possible targets for therapeutic intervention.

Head and neck cancer is the seventh most prevalent cancer worldwide accounting for nearly 900,000 new cases annually. The vast majority of cases are squamous cell carcinomas (SCC) including its subtypes and are located in the oral cavity (OSCC), pharynx and hypopharynx, larynx, nasal cavity and its sinuses, and ear [15]. Nearly all cases of SCC in the salivary glands are metastatic from cutaneous SCC [15]. Here we present a brief update on epigenetic alterations in these cancers (Table 1).

Table 1.

Summary of important gene methylations and their utility

Location & tumour type	Genes	Utility value
		Prognostic	Diagnostic
Oral SCC	CDKN2A	Yes	–
	DAPK	Yes	–
	TIMP3	Yes	–
	MGMT	Yes	–
	KDM1A	Yes	–
Nasopharyngeal carcinoma	RASSF1	Yes	Yes
Sinonasal malignancies
Adenocarcinoma (intestinal type, ITAC)	TIMP3	Yes	–
NUT	RASSF1
	BRD4 – NUT	Yes	–
	(fusion)/EZH2	Yes	Yes
SWI/SNF complex-deficient carcinoma	SMARCB1	Yes	–
	SMARCA4	Yes	–
	RASSF1	Yes	–
SNUC	IDH2	Yes	–
Olfactory neuroblastoma	EDH2, IDH2	Yes	(Yes)
Salivary gland carcinoma
MEC, ADCC, ACC, CXPA	RASSF1	Yes	–
MEC, ADCC	TIMP3	Yes	–
MEC, CXPA	CDKN2A	Yes	–
SDC	ERBB2	Yes	Yes

SCC, OSCC and nasopharyngeal carcinoma

A large study of head and neck squamous cell carcinoma (HNSCC) identified 27 aberrantly methylated genes and showed that FAM135B methylation (family with sequence similarity 135 member B) was an independent prognostic biomarker for overall survival [16]. The CDKN2A gene (cyclin dependant kinase inhibitor 2 A) provides instructions for making the p16(INK4A) and p14(ARF) proteins, both of which function as tumour suppressors. In head and neck cancer CDKN2A is one of the most well-studied TSGs with regard to methylation and epigenetic alterations. Mutations in this gene are estimated to be found in 25% of HNSCC. A relatively recent systematic literature review of the epigenetics of head and neck cancers included 25 studies with a total number of 3,790 samples of OSCC and oral potentially malignant disorders. DNA methylation was reviewed in 18 studies and the role of histone modifications was assessed in seven studies. Histone H3 modification, along with DAPK (death-associated protein kinase), TIMP3 (tissue inhibitor of metalloproteinase 3), and CDKN2A were the most commonly investigated epigenetic biomarkers. Assessment of the prognostic value of the epigenetic biomarkers was not performed in any of these studies and the clinical applicability of the studies is limited [17, 18]. There are, however, promising ongoing clinical trials such as treatment with low-dose 5-azacytidine.The use of DNMTi 5-Aza for patients with HNSCC sensitizes anti-PD-L1 refractory cancers to reprogramme host immune responses to increase PD-L1 and IFN-γ expression and thereby prolonging the overall survival [19]. In a large review study, the impact of DNA methylation on the risk of oral cancer development was analysed. Promoter methylation of MGMT (methylguanine methyltransferase), DAPK, and CDKN2A displayed a significant prognostic value and are hence promising biomarkers for oral cancer [20]. Lysine-specific demethylase, LSD1 (encoded by the KDM1A gene), is a histone demethylase that is important by promoting cancer initiation as well as cancer progression. Several studies from Manish Bais’s laboratory in Boston have demonstrated the crucial role of LSD1 protein and that inhibition of LSD1 epigenetically attenuates oral cancer development and growth [21–24]. Epigenetic alterations in oropharyngeal SCC reveal distinct differences between HPV-positive and HPV-negative SCC [25, 26]. HPV-negative cancers have a considerably higher degree of genome-wide hypomethylation suggesting that they are far more genomically unstable than HPV-positive cancers which would account for deregulation of cellular processes typical of aggressive neoplasms [27]. Smoking and alcohol have widespread dose dependant epigenetic alterations, primarily via DNA methylation but also histone acetylation, reflecting accelerated biological ageing. Nasopharyngeal carcinoma (NPC) is a rare malignancy in most parts of the world but with a remarkable ethnic and geographic distribution in southern China, Southeast Asia, North Africa, and the Arctic/Alaska. Non-keratinizing squamous cell carcinoma (NK-NPC) is the most common of the three histological subtypes of NPC. The vast majority of NPC are Epstein-Barr Virus (EBV)–associated and EBV acts as an epigenetic driver during NPC pathogenesis [28]. The DNA methylation status of the TSG RASSF1 can be used as a complement in NPC diagnostics [29]. EBV DNA methylation can also be used in distinguishing NPC from nasal NK/T-cell lymphoma [30].

Sinonasal malignancies

Newly defined and emerging sinonasal carcinoma entities have different histologies and genomic profiles compared to conventional non-keratinizing and keratinizing SCC, and adenocarcinoma. Epigenetic events in sinonasal SCC and sinonasal adenocarcinoma have been studied well before the new entities were defined. Sinonasal SCC has fewer gene methylation events compared to adenocarcinoma (intestinal-type, ITAC). RASSF1, TIMP3, and CDH13 (Cadherin 13) are methylated in both types of carcinomas but have no correlation to clinicopathological features or overall survival in SCCs, contrary to ITAC where TIMP3 methylation is associated with a significantly worse survival [31].

NUT carcinoma (Nuclear Protein in Testis), SWI/SNF (SWItch/Sucrose Non Fermentable) complex-deficient sinonasal carcinoma (also termed SMARCB1- and SMARCA4-deficient undifferentiated sinonasal carcinoma), and SNUC (Sinonasal Undifferentiated Carcinoma) constitute new carcinoma entities that require the use of molecular identification, mostly detectable by, surrogate antibodies, for classification. SNUC may still be best regarded as a diagnosis of exclusion [32–35]. A fair number of epigenetic studies have been performed on undifferentiated sinonasal carcinomas. Olfactory neuroblastoma (ONB) is a rare sinonasal malignancy and is composed of various cell lineage-specific elements from the olfactory epithelium. Currently there are limited therapeutic options for these cancers. Epigenetic data is relatively limited but methylation profiling of a series of 66 cases of ONB distinguished four unique subgroups. The largest group (64%) had global hypermethylation and one of the smaller groups had a CpG island methylator phenotype (CIMP) and also IDH2 (Isocitrate Dehydrogenase 2) hotspot mutations [36]. Drugs targeting the mutated forms of IDH2 are in use for treatment of some cancers associated with these mutations. A relatively recent literature review (articles before 2020) of the genomics and epigenetics of ONB was published in 2021 [37]. The Polycomb group enzyme EZH2 (Enhancer of Zeste 2) is a critical component of a multiprotein complex with an important role in catalysing the trimethylation of lysine 27 on histone H3 which leads to silencing of many genes, including TSGs. The EZH2 methyltransferase activity in metastatic ONB has been suggested to be higher than that found in primary ONB. ONB tumours with low EZH2 repression signature and high EZH2 activity may be susceptible to treatment with EZH2 inhibitors, possibly increasing the susceptibility to immunotherapy [38, 39].

SWI/SNF complex-deficient carcinomas have loss of one SWI/SNF complex gene, SMARCB1 and SMARCA4 being the most relevant in head and neck cancer [40]. A statistically significant higher methylation of the TSG RASSF1 (Ras association domain family member 1) has been demonstrated in SMARCB1/INI1-deficient compared to SMARCB1/INI1-positive tumours [41]. RASSF1 methylation has also been demonstrated in a study of NUT carcinoma where all other 23 genes investigated were unmethylated [42]. Studies of 25 TSGs in sinonasal carcinomas revealed significantly higher DNA methylation in GATA5, THSB1 (encodes the thrombospondin-1 protein), and PAX5 genes, and methylation in five or more genes was associated with impaired overall survival [43]. At least two excellent studies on potential DNA methylation-based classification of sinonasal undifferentiated carcinomas have been proposed. IDH2 mutations in sinonasal carcinomas induce a hypermethylator phenotype (SNUC and large cell neuroendocrine carcinoma) distinct from small cell neuroendocrine carcinoma and SMARCB1-deficient sinonasal carcinomas. Tumours with SNUC morphology may not be as undifferentiated as previously considered and may be divided into at least three classes, possibly four. Two of these (having IDH2 or SMARCA4/ARID1A mutations) with more favourable clinical course than the highly aggressive SMARCB1-deficient carcinomas [44, 45].

Salivary gland cancer

This is a very heterogeneous group of neoplasms, albeit being rare, and where the malignant epithelial tumours currently are histologically classified into 21 different subtypes. They often have overlapping histologies, different genomic profiles and biological behaviour. Due to their rarity and diversity personalized treatment can be difficult to evaluate. The important role of epigenetic alterations in cancer initiation and development has been recognised for several decades and some initial studies were performed already 20 years ago. These studies included for example promoter methylation of CDKN2A, RASSF1A and DAPK in adenoid cystic carcinoma (ADCC). A high frequency of RASSF1A methylation was noticed in high-grade tumours and in tumours with metastasis, suggesting that methylation of this gene may play an important role in the progression of ADCC [46–48]. Mucoepidermoid carcinoma (MEC) showed hypomethylation of CLIC3 (coding chloride intracellular channel protein 3) but no correlation with clinical or pathological characteristics [49]. Another study investigated the prevalence of promotor methylation of CDKN2A, TP53 and TERT genes in MEC. The highest value was for CDKN2A and interestingly TERT was more often methylated in lower clinical stages [49`]. Two comprehensive reviews on epigenetic alterations in salivary gland cancer were published 3–5 years ago summarizing the methylated genes in the most common salivary carcinomas. For example, RASSF1A was methylated in MEC, ADCC and acinic cell carcinoma (ACC), and in carcinoma-ex-pleomorphic adenoma (CXPA), TIMP3 in MEC and ADCC, and CDKN2A in MEC and CXPA [50, 51]. In a review of 47 studies of different salivary gland tumours (SGT) meta-analysis revealed that both MGMT (methylguanine methyltransferase) and RASSF1A were methylated and related to higher stages of the tumours but not associated with the grade of the tumours [52]. A comprehensive study of 363 cases encompassing 20 SGT entities provided a DNA methylation landscape of SGTs. Most entities had specific epigenetic signatures, cribriform adenocarcinoma being epigenetically distinct from polymorphous adenocarcinoma and myoepithelioma and pleomorphic adenoma (PA) formed their own unique class [53]. There is currently insufficient evidence to prove a link between SGTs, HPV infection and DNA methylation of TSGs [54]. Studies of TERT promoter (THOR) methylation in parotid PA, recurrent PA, CXPA and in adjacent histologically normal parotid tissue indicate that THOR methylation in recurrent PA is closer to carcinoma than to normal tissue. A subset of PA-adjacent tissues showed epigenetic alterations, possibly suggestive of increased risk of recurrence [55]. Another recent study used DNA methylation profiling for differentiating between PA and CXPA. Three distinct clusters, benign, intermediate and malignant, were identified. Pathogenic mutations of TP53, HRAS, PTEN and/or TERT were only present in CXPA [56].

Summary

Epigenetics is an evolving science within the biomedical field. The main difference between the molecular pathology of gene changes and epigenetic alterations is that the latter are reversible offering new therapeutic possibilities. Methylation profiling can be used as a complement in for example distinguishing adamantinoma-like Ewing sarcoma from conventional Ewing sarcoma [57] and nasopharyngeal carcinoma from nasal NK/T cell lymphoma [30]. Within head and neck cancer, a number of recurrent epigenetic alterations with clinical implications both with regard to diagnosis, prognosis, and therapy, have recently been demonstrated and several clinical trials are now ongoing.

Author contribution

Conceptualization and writing the original draft preparation: H. Hellquist, P. Castelo-Branco – review and editing: G. Stenman, A. Nadal, A. Agaimy, N. Zidar, RV de Lima-Souza, F.V. Mariano, A. Coca-Pelaz A, and A. Ferlito. All the authors have read and agreed to the submitted version of the manuscript.

Funding

Open access funding provided by FCT|FCCN (b-on).

Data availability

Data supporting the findings of this study are available within the article.

Declarations

Conflict of interest

Abbas Agaimy is Editor-in-Chief of Virchows Archiv. Other authors declare no competing interests.