Molecular biology in conjunctival melanoma and the relationship to mucosal melanoma

Conjunctival melanoma is a rare malignancy of melanocytes located in the conjunctiva – the mucosal membrane covering the eye. The course of disease is often uncomplicated with proper treatment; however, a subset of patients develop metastases that can be difficult to treat and are associated with a poor prognosis. Unlike all other mucosal melanoma, conjunctival melanomas are partly exposed to sunlight, similar to cutaneous melanoma. Previous studies of mucosal melanomas have not included conjunctival melanomas, and studies of conjunctival melanomas have mostly been compared with cutaneous melanoma. Therefore, the relationship between conjunctival melanoma and other mucosal melanoma remains fairly unknown. In the era of precision medicine, prognostic markers and therapeutic targets remain of high importance in cancer management. However, in conjunctival melanoma, no markers for metastatic disease exist at the moment. In order to contextualize conjunctival melanoma as a mucosal melanoma, we conducted a series of studies, including a literature review investigating the clinical and molecular characteristics of conjunctival melanoma (I) and an original study investigating the genetic and molecular profile in conjunctival and other mucosal melanomas (II). Furthermore, we investigated whether miRNA could be used as a prognostic marker for metastasis in conjunctival melanoma (III). We identified conjunctival melanoma as being the most frequent mucosal melanoma with a fairly good prognosis overall. We found that conjunctival melanoma has a distinct molecular profile when compared to other mucosal melanomas. This difference was highlighted by frequent mutations in BRAF and a differential expression of 41 genes. Interestingly, sun exposure could only in part explain this difference at the gene expression level. When investigating miRNAs as potential prognostic markers, we observed that primary conjunctival melanomas with and without metastatic potential separated at the miRNA level; however, we were not able to identify one specific miRNA that was predictive for metastatic disease. In conclusion, conjunctival melanoma is a special variant of mucosal melanoma that shares only some characteristics with cutaneous melanoma. miRNA could not be confirmed to be a marker of metastases, and future studies of multiple markers are needed in order to predict metastatic behaviour. Conjunctival melanoma is a rare cancer that mainly affects elderly Caucasians (Larsen et al. 2015). Melanoma is a malignant transformation of melanocytes, and consequently, melanoma may be found at all anatomical locations where melanocytes are present (Mikkelsen et al. 2016). These locations include skin melanoma (91%), ocular melanoma (uveal tract and conjunctiva) (5%), other mucosal melanoma (1%), and melanomas with unknown site of primary tumour (3%) (Chang et al. 1998). Of the ocular melanomas, 85% are constituted by melanomas of the uveal tract and 15% are conjunctival melanomas (Osterlind 1987; Chang et al. 1998; Seregard 1998; McLaughlin et al. 2005). Being in fact a mucosal melanoma, a special feature of conjunctival melanoma is their exposure to sunlight. Most conjunctival melanomas are located in the sun-exposed bulbar conjunctiva, and conjunctival melanoma located in the fornical and tarsal conjunctiva is generally not considered to be sun exposed (Mikkelsen & Heegaard 2018). In contrast, mucosal melanomas of other sites are completely sun-shielded. Interestingly, extrabulbar conjunctival melanoma has been shown to be associated with a poor prognosis more similar to that of mucosal melanomas from non-sun-exposed locations (Larsen et al. 2015; Mikkelsen et al. 2016). Whether this is due to a delayed diagnosis or a molecular similarity to other mucosal melanomas that generally carry a grave prognosis is unknown; however, studies of mutations in the BRAF gene suggest the latter being the case (Larsen et al., 2016a,b; Mikkelsen & Heegaard 2018). In recent years, the molecular biology of cutaneous melanocytic lesions and in particular cutaneous melanoma have been studied in great detail. In particular, a great deal of work has been done in exploring the genetic landscape of cutaneous melanoma, which is clearly reflected in the recent World Health Organization (WHO) classification of skin tumours fourth edition (2018). In this volume, malignant melanoma is separated into different subgroups according to their pathway from melanocyte to malignant melanoma, based on genetic profile and precursor lesion. Regarding conjunctival melanoma, only few molecular studies have been carried out. Mutations in the BRAF, NRAS, and NF1 genes have been identified (Spendlove et al. 2004; Beadling et al. 2008; Wallander et al., 2011; Alessandrini et al. 2013; Griewank et al. 2013a,b; Sheng et al. 2015; Larsen et al., 2016a,b; Scholz et al. 2018), but still, none of these seem to correlate with the course of disease or prognosis (Larsen 2016). In particular, the metastatic pattern is puzzling – with all from multiple local recurrences to very late occurring distant metastases being seen (Larsen 2016). In this series of studies, we investigated conjunctival melanoma at different molecular levels, including genetics, gene expression, and epigenetics, in order to provide a piece of the conjunctiva melanoma puzzle. The conjunctiva is innervated by the ophthalmic division of the trigeminal nerve (V1), mainly through branches from the nasociliary nerves, lacrimal nerve, frontal nerve, and the infraorbital nerve. Close to the limbus, the mucous membrane is innervated by the ciliary nerves (Duke-Elder & Wybar 1961). The arterial supply is provided from the palpebral branches of the nasal- and lacrimal arteries of the lids and from the anterior ciliary arteries. The venous drainage from the conjunctiva is to the anterior ciliary vein along with the superior and inferior ophthalmic veins. The conjunctival lymphatic vessels form an irregular mesh under the epithelium and drain to the superficial parotid (pre-auricular), submandibular, and deep cervical lymph nodes (Duke-Elder & Wybar 1961). Microscopically, the normal conjunctiva consists of two histological layers, namely the epithelium and the underlying lamina/substantia propria. The epithelium is mainly composed of stratified columnar epithelium; however, in the transition zones close to the lid margin and the corneal limbus, the conjunctival epithelium transforms to non-keratinizing stratified squamous epithelium. Numerous mucous-secreting goblet cells and melanocytes are present within the epithelium. The lamina propria may be subdivided into two histological layers: a superficial adenoid layer and an underlying fibrous layer. The adenoid layer consists of loose connective tissue that contains lymphocytes, mast cells, dendritic cells, and macrophages. This lymphoid tissue is known as mucosa-associated lymphoid tissue (MALT) (Knop & Knop 2000), in which most conjunctival lymphomas arise (Mikkelsen et al. 2018). The deep fibrous layer consists of dense connective tissue in which collagen and elastic fibres are found. This layer is more properly considered a subconjunctival connective tissue layer, and in it lie the vessels and nerves supplying the conjunctiva (Duke-Elder & Wybar 1961). Cells in our body are derived from one of the three germ layers known as the ectoderm, the endoderm, and the mesoderm. The ectoderm can be further subdivided into the surface ectoderm and the neuroectoderm – the latter giving rise to the neural crest and the neural tube. The conjunctiva is derived from the surface ectoderm, which also gives rise to other mucosal membranes of the head and neck, such as the surface epithelium of the oral and nasal cavities (Duke-Elder & Cook 1963). In the sixth week of gestation, two small folds of surface ectoderm (the frontonasal process and the maxillary process) become visible cranially and caudally to the developing cornea. These two processes give rise to the eyelids, and they grow towards each other to merge in the eighth week of gestation. Between the mesodermal developing cornea and the merged early eyelids, a space is formed called the conjunctival sac. The conjunctival sac is visible from around the tenth week of gestation, and, inside this space, the conjunctival mucous membrane develops. Between the fifth and seventh month of gestation, the eyelids separate again, and the fornical conjunctival folds become visible in the last month of gestation (Duke-Elder & Cook 1963). The function of the conjunctiva is to provide a frictionless surface that allows for free and smooth eye movement. Furthermore, the conjunctiva plays an important role in tear film stability, corneal transparency, and in protecting the eye and orbit from infection (Seregard et al. 2015). Melanocytes are the pigment-producing cells found in the skin, hair, and eyes (Busam et al. 2014). The pigment produced by melanocytes is called melanin, and, in the skin, melanin has great importance in protection from ultraviolet (UV) radiation from the sun by forming an intracellular cap, protecting the nucleus of the surrounding epithelial cells (Cichorek et al. 2013; Busam et al. 2014; Mikkelsen et al. 2016). Melanocytes can also be found in most mucosal membranes, where their role is still not understood (Mikkelsen et al. 2016). Melanocytes develop from melanoblasts that are generally thought to originate in the neural crest, a part of the neuroectoderm (Cichorek et al. 2013). The neural crest harbours multipotent stem cells, which can develop into neurons, Schwann cells, other glial cells, chondrocytes, muscle cells, endocrinal cells, and melanocytes (Hirobe 2011). From the neural crest, melanoblasts migrate dorsolaterally between the ectoderm and the somites to their definitive sites of colonization, following the route of other neural structures, such as peripheral nerves (Thomas & Erickson 2008; Hirobe 2011; Cichorek et al. 2013; Agarwalla et al. 2015). Notably, a subpopulation of melanoblasts follow an alternative ventral route (Cichorek et al. 2013). During the migration from the neural crest, the multipotent melanoblast gradually matures, and in the case of skin melanocytes, the final maturation into melanocytes takes place within the dermis (Hirobe 2011; Cichorek et al. 2013; Busam et al. 2014). This process may be the same in the conjunctiva; however, this is not known. The maturation process is regulated by the transcription factor Sox-10, which activates the microphthalmia transcription factor (MITF), which is the major driver of the melanocytic phenotype (Busam et al. 2014). Interestingly, experiments have shown that terminally differentiated melanocytes are able, under certain conditions, to dedifferentiate, self-renew, and even differentiate to other neural crest-derived cells such as Schwann cells (Dupin et al. 2000). This is thought to be a common feature of neural crest cells, and, seen in this light, it is not surprising that other neural crest-derived tumours, such as schwannomas, may also present a pigmented phenotype (Agarwalla et al. 2015). Microscopically, melanocytes are small round or oval cells with clear cytoplasm in hematoxylin and eosin (H&E)-stained sections. In the skin and conjunctiva, the melanocytes are distributed territorially, with a regular distance within and just beneath the basal epithelial cell layer. Variable numbers of melanocytes are seen in the basal layer of the perilimbal conjunctiva (Spencer 1996). Melanocytes have several dendrites that function as intercellular contacts between the melanocyte and several basal epithelial cells. Ultrastructurally, melanocytes are characterized by a prominent Golgi complex, rough endoplasmatic reticulum, numerous mitochondria, and the presence of melanosomes, the unique ultrastructural hallmark of melanin-producing cells. Melanosomes contain enzymes involved in melanogenesis (Busam et al. 2014). The concept of cancer being a disease of the genome where a single cell overcomes the inherent suppression of cell division due to changes in the chromosomes was first proposed in 1914 by Theodor Boveri (1862–1915) (Manchester 1995). In 1958, Francis Crick (1916–2004) initially described the "central dogma of molecular biology" as the process of the transfer of genetic information from nucleic acid (DNA) to nucleic acid (RNA) to protein, for which he was awarded the Nobel Prize in medicine in 1962 (Crick 1958). Since the early 1970s, molecular oncology research has focused on identifying genes that are frequently mutated in cancer, and it is now generally accepted that cancer arises through the accumulation of mutations in genes (oncogenes) recognized as either proto-oncogenes or tumour suppressor genes. Proto-oncogenes are defined as normal genes that, if overexpressed or constitutively active due to a mutation, can contribute to cancer formation. A myriad of such genes has been identified and includes BRAF, NRAS, and MYC. On the other hand, tumour suppressor genes are genes that protect the cell from becoming cancerous, and when underexpressed or mutated, causing a loss or decrease of function, the cell is likely to progress to cancer. Examples of such genes include NF1, TP53, and PTEN. Recently, a third class named "stability genes" has been identified. Mutations in these genes result in increased mutational rate throughout the genome, which increases the risk of mutations in proto-oncogenes and tumour suppressor genes (Vogelstein & Kinzler 2004). Regardless of the location of genetic alterations in either proto-oncogenes or tumour suppressor genes, the malignant phenotype is characterized by various cellular functions, conceptualized by Hanahan and Weinberg in 2000 as the "hallmarks of cancer" (Hanahan & Weinberg 2000; Hanahan & Weinberg 2011). These hallmarks define eight characteristic features of cancer cells, namely (i) sustained proliferative signalling, (ii) evasion from growth suppressors, (iii) evasion from apoptosis, (iv) limitless replicative potential, (v) sustained angiogenesis, (vi) tissue invasion and metastasis, (vii) altered energy metabolism, and (viii) evasion from immune destruction (Hanahan & Weinberg 2000; Hanahan & Weinberg 2011). In melanoma, genetic alterations disrupt a common set of key signalling pathways, and in particular the mitogen-activated protein kinase (MAPK) pathway is affected in most melanomas at an early stage (Figure 2) (Bastian 2014). The MAPK pathway is activated by mutations at several levels in the pathway. First, activating mutations are found in the upstream surface receptor at the level of KIT. Secondly, mutations are seen in RAS family members acting at the level directly downstream of surface receptors along with negative regulators of RAS (e.g. NF1 and SPRED1). Thirdly, mutations can be found in RAS effectors acting at the levels further downstream (e.g. BRAF, MEK, and cyclin D1) (Figure 2) (Dahl & Guldberg 2007; Zhang et al., 2016; Bastian et al., 2018). In untreated melanomas, alterations of these driver genes tend to be mutually exclusive; however, they can co-occur in melanomas following targeted therapy and contribute to resistance of therapy (Bastian et al., 2018). Following the initial activation of MAPK by early mutations in driver genes in melanoma, a myriad of secondary mutations and alterations occur (Bastian 2014; Shain et al. 2015). A common target for these secondary mutations is the G1/S checkpoint, and mutations affecting this checkpoint allow the cell to override the ability to restrain cell-cycle entry (Bastian 2014). Furthermore, mutations affecting pathways related to the SWI/SNF chromatin remodelling complex are often found, including inactivations of the histone modifier BAP1, which is often mutated in uveal melanoma (Harbour et al. 2010; Smit et al. 2018). Other secondary alterations in melanoma include deletions or mutations in CDKN2A (encoding p16), amplification or mutation of CDK4 (encoding the target kinase of p16), along with loss of RB1. Several of these events can exist as germline alterations predisposing to melanoma. The most frequent mutations in cutaneous melanoma are found in the promotor region of TERT, resulting in immortalization of the cell due to increased expression of telomerase. The mechanisms leading to mutations in driver genes of melanoma vary considerably depending on the type of melanoma (Bastian 2014). The genome of skin melanoma often carries numerous mutations related to ultraviolet light (UV) exposure, mostly cytosine to thymidine (C>T) transitions (Trucco et al. 2019). Melanomas occurring in skin with some degree of cumulative sun damage (CSD) are among the most highly mutated cancers overall, harbouring approximately 30 mutations/Mb of DNA, which is equal to 100 000 mutations pr. genome (Hayward et al. 2017). Skin melanomas with a low degree of CSD (non-CSD) only have about half the mutational load (e.g. 15 mutations/Mb DNA). In contrast, acral melanoma and mucosal melanoma have considerably lower mutational burdens and typically lack a UV signature. Instead, these melanoma genomes are characterized by numerous copy-number variations (CNV) and structural variants (SV). In cutaneous melanoma, atypical mutations in BRAF are found in CSD melanoma, while the BRAF-V600E mutation is associated with non-CSD melanoma and development from an acquired nevus (Shain et al. 2015). Uveal melanoma stands out as a separate melanoma subtype derived from non-epithelial melanocytes resembling melanomas occurring in blue nevi and internal organs, all characterized by Gαq signalling. These genomes are characterized by a lack of highly rearranged genomes along with a low mutational burden, mostly with primary mutations in GNAQ or GNA11 followed by few progression mutations in SF3B1, EIF1AX, and BAP1 (van Poppelen et al., 2018; Shain et al. 2019). Genetics have been shown to play a central role in cancer starting in a normal cell that progresses to a premalignant lesion and further develops into malignancy and metastases. In recent years, we have learned much about the important roles of certain genes and mutations in cancer, but the inadequacy of genetics alone in explaining cancer biology has been known for many years. Gene expression is the process transferring genetic information to a functional gene product, and this process contains many steps, including transcription, RNA splicing, translation, and post-transcriptional modification of the gene product (most often a protein). Several mechanisms contributing to carcinogenesis and cancer biology by altering the gene expression levels have been recognized, including a spectrum of epigenetic mechanisms such as gene-silencing by methylation, modification of chromatin structures, and post-transcriptional gene regulation. Ultimately, disturbances of the gene expression lead to altered signalling of the affected genes, which causes a dysregulation of various signalling pathways. In cancers, such as breast cancer, cutaneousmelanoma, and diffuse large B-cell lymphoma (DLBCL), gene expression profiling has revealed several subtypes that correlate with prognosis, course of disease, and response to certain treatments (Alizadeh et al. 2000; Perou et al. 2000; Mikkelsen et al. 2018). In conjunctival melanoma, no profiling of cancer gene expression has been carried out. However, several studies have investigated the expression of a few or single targets, such as KIT and beta-catenin (Beadling et al. 2008; Reddy et al. 2017). Recently, a sophisticated study from the Netherlands showed that enhancer of zeste homolog 2 (EZH2) was highly expressed in conjunctival melanoma without having concurrent mutations in the gene (Cao et al. 2018). miRNAs are small non-coding RNA molecules (~22 that can gene expression at the post-transcriptional level. is recognized as an epigenetic and can be by the miRNA according to one of the following mechanisms et al. et al. of the resulting in of the resulting in of the or of resulting in miRNAs with the complex and form the complex that the gene The miRNA is to a region of the and it that one miRNA may have several target target the region (the first of the has to with the In this target is often at the of the and the of the miRNA is often et al. et al. 2014). In cancer, of a miRNA a tumour suppressor gene to of the gene leading to of the tumour suppressor gene and of tumour formation. In contrast, of a miRNA an to of the leading to expression and increased tumour et al. et al. 2014). However, miRNAs most often have multiple targets for their ability to contribute to several cellular pathways, the function of specific miRNAs In some cancers, such as breast cancer, miRNAs can be used as prognostic markers et al. 2019). In conjunctival melanoma, a study by our identified and miRNAs in conjunctival melanoma (Larsen et al., Several of these miRNAs have been identified in cutaneous melanoma, and the study an entry for miRNA in conjunctival melanoma as potential prognostic and therapeutic The first of a conjunctival malignant melanoma in the literature was by the in A of the of the and their from which is also recognized as the first in In this a case of conjunctival a of an in an cornea and However, the first of a of the conjunctiva was by the in In the following numerous similar were by different malignant or most frequently (Duke-Elder & was the used by and in when they the first review of the including & In used the and this was the first the was used for this in the literature in of from that most of these were the same disease, namely malignant melanoma (Duke-Elder Conjunctival melanoma as a separate in the first edition of the of eye known as and melanoma was also the used by in when he described late conjunctival melanoma metastases & is that described local and metastases as features in conjunctival melanoma in These are still for ocular in the era of medicine, and molecular biology may in of metastasis that of these The of conjunctival melanoma is in (Seregard & et al. Seregard 1998; et al., et al. et al., Larsen 2016; et al. 2018; et al. 2019). This is considered a high rate when the surface to that of the skin and other mucosal Conjunctival melanoma in Caucasians and is very rare among and et al. In a recent study from the has been to be et al. 2015). During the the rate has been in the et al. however, recent studies from and all an in in recent et al., et al. Larsen 2016). In this is most located in the sun-exposed conjunctiva, that ultraviolet radiation may play a role in the (Larsen et al., The in is also found in skin melanoma but is in uveal melanoma et al. The of conjunctival melanoma among and is considered to be however, studies from and have towards a et al., et al. et al. 2019). Furthermore, some studies have patients to be at the of diagnosis et al. et al., et al. et al. 2011). Most patients with conjunctival melanoma are with a of years, and are only affected et al. et al., et al. an & 2007; et al. 2008; et al. Conjunctival melanoma arises from melanocytes present in the conjunctiva, and are three recognized pathways leading from a melanocyte to a conjunctival melanoma. A conjunctival melanoma can develop from a melanocytic such as primary acquired from a or it can develop from a single melanocyte located in the conjunctiva without present premalignant lesion The most common precursor lesion for conjunctival melanoma is which has been to be present in of all of conjunctival melanoma, most of these occurring in Caucasians et al. et al. et al., et al. & et al. Larsen et al. 2015). acquired is a as not all are considered the included a clinical and a leading to much is characterized as a and lesion with of located in the conjunctiva with or without to the skin cornea et al., Seregard 1998; & Larsen 2016). However, can be further into with cellular which the premalignant melanoma in of skin melanoma to various and without cellular which is recognized as a lesion without malignant potential et al., et al. 2018). Whether a is as or has been associated with some degree of and and as an to the and a was proposed a few the conjunctival melanocytic & In to the of a to the changes associated with is seen as a of the The from to where is without and melanoma in & Conjunctival melanocytic lesion is the in the recent classification that a of changes from melanocytic to melanoma in (2018). These three are being by a large conjunctival nevi progress into melanoma & of conjunctival melanomas arise in a nevus et al. et al. et al., et al. et al. 2011; Larsen et al., conjunctival nevi are characterized by melanocytes nevus located the epithelium that form characteristic et al. can be further according to the location of the nevus in the conjunctiva et al. Seregard et al. 2015). are typically formed in the first of where the nevus can be found in the just beneath the epithelium et al. Seregard et al. 2015). the nevus into the propria