Background
Primary mucosal melanoma (MM) is a rare subtype of melanoma which accounts for approximately 1% of melanoma and arises from melanocytes in mucosal tissue of different anatomical sites, such as the head & neck, gastrointestinal tract, or genitourinary tracts [
1‐
3]. Compared to cutaneous melanoma (CM), MM is relatively asymptomatic or lacks early clinical visibility, which is important for early detection of CM, thus it is often diagnosed at more advanced stages and therefore, exhibiting poor prognosis. The treatment of MM remains subjective because of the rareness of the cancer and lack of randomized controlled trials [
1‐
3]. Epidemiologic studies have indicated potential risk factors for MM, such as tobacco exposure, HIV infection, or chronic inflammation [
4], however precise roles of these factors on MM remain unknown.
In a recent study [
5], CM was stratified into four molecular subtypes:
BRAF mutated,
RAS mutated (
NRAS/
KRAS/
HRAS),
NF1 mutated (a regulator of RAS pathway [
6]), and triple wild-type (Triple-WT, a subgroup that lacks above mutations) using the Cancer Genome Atlas (TCGA) database. Molecular targeted therapies, such as BRAF inhibitors or MEK inhibitors, are applicable for CM treatment based on these genetic subtypes [
7]. However, mutational patterns in MM have been profiled for only a few genes, such as
BRAF,
NRAS, and
KIT, which are mostly targeted for specific hotspots or limited regions within the genes [
1,
3,
8]. Our knowledge about the cancer-related gene mutations in MM, particularly in all exonic regions, is still limited and warrants further investigation into the mutational landscape to understand the etiology of the disease and better treatment strategies for MM.
To date, the majority of genomic studies aimed to identify somatic mutations in melanoma have focused mainly on CM and have only include a small number of MM samples (i.e. [
9,
10]). Therefore, to better investigate the etiology of MM and to clarify mutations that are potentially relevant for MM treatment, we utilized targeted next-generation sequencing (NGS), which enabled us to screen all exons on multiple cancer-related genes with high coverage [
11‐
14] in a large collection of MM. Here, we analyzed the mutational landscape between 41 MM and 48 CM specimens using our custom panel of 89 genes frequently mutated in cancer.
Discussion
This study identified distinct mutational landscapes between MM and CM, particularly the signature for UV-induced DNA damage, and revealed that common driver gene mutations for CM were less frequent in MM. Although malignant transformation of melanocytes into CM is highly related to UV damage, and
BRAF or
NRAS mutations are involved in the progression of CM [
5,
34], our study strongly suggested that MM likely has distinct mechanisms involved in its initiation and progression pathways. This is presumably from tobacco exposure or mutation in
IGF2R and
KIT, and thus MM may require different treatment strategies from CM.
Triple-WT comprises 15% of CM, as was previously identified in the melanoma TCGA cohort [
5]. This subtype lacks hotspot
BRAF,
RAS, or
NF1 mutations which are important driver genes for CM. Triple-WT in CM has unique molecular characteristics such as amplifications of KIT, PDGFRA, VEGFR2, MDM2, or TERT, as well as an enrichment of complex structural rearrangements like fusion of driver genes [
5]. Compared to CM, MM was highly associated with Triple-WT (70.7%). Despite the distinct characteristics between Triple-WT and non-Triple-WT, we observed no significant difference in mutational spectrum, tumor mutation burden, or prognosis between Triple-WT and non-Triple-WT in MM. Further genetic and epigenetic landscapes need to be elucidated to comprehensively investigate biological and clinical relevance of Triple-WT in MM.
In addition to Triple-WT, our study provides several important implications for the treatment of MM, particularly related to mutations in
IGF2R and
DCC genes. MM was significantly associated with
IGF2R mutations, which are relatively low in other cancer types from TCGA database (Fig.
3b), indicating the unique genetic background of MM. Notably, none of the
IGF2R mutations were recurrent at a single locus, signifying the importance of screening all exons within the gene panel. IGF2R is a multifunctional receptor and is involved in the IGF pathway [
42]. The IGF pathway is triggered by IGF ligands (insulin, IGF1, or IGF2) binding to their receptors (insulin receptor or IGF1R) [
46]. Stimulation of the pathway contributes to carcinogenesis or tumor progression in different tumors, including melanoma [
40,
46‐
48]. IGF2R also has a high affinity for IGF ligands, particularly IGF2; however, the receptor lacks an intracellular tyrosine kinase domain that is essential for the activation of the IGF pathway, thus, the receptor acts as a “decoy” of the IGF pathway and is recognized as a tumor suppressor gene [
42,
46,
47]. Although deregulation of IGF pathway through amplification or overexpression of IGF2 is involved in another mucosal-origin tumor, CRC [
17], clinical relevance of
IGF2R mutations is still controversial [
42]. Inhibitors targeting the IGF pathway, such as anti-IGF1R antibodies in ongoing clinical trials [
49], are potential candidates for MM treatment.
DCC mutations is another candidate that is potentially relevant for MM treatment.
DCC codes netrin-1 receptor, which prevents apoptosis by binding to netrin-1. However, netrin-1 shortage induces cleavage of DCC at its intracellular domain and promotes apoptosis; thus,
DCC is considered as a tumor suppressor [
50‐
52]. Although the significance of
DCC in melanoma remains unknown, mutations could lead to deregulation of DCC, possibly affecting the prognosis of MM. Interestingly,
DCC mutations were more prevalent in CM; however, it was significantly associated with poor prognosis in MM, but not in CM. These results imply distinct biological significance of
DCC mutations in MM and CM.
In addition to molecular targeted therapies applicable for individual mutations, immune checkpoint inhibitors that target CTLA-4 (cytotoxic T lymphocyte antigen 4), PD-1 (Programmed death 1), or PD-L1 (Programmed death ligand 1) demonstrates great promise for treatment of different tumors, including CM [
39‐
41]. Particularly, drugs that block PD-1 (nivolumab or pembrolizumab) lead to significant improvement in CM treatment [
40,
41]. MM also demonstrates higher response to nivolumab compared to ipilimumab (CTLA-4 inhibitor) [
53]. PD-L1 expression, which is positively associated with tumor mutation burden [
38], is clinically important as it predicts a better response to anti-PD-1 therapy in CM [
40,
41]. In this study, MM demonstrated significantly lower tumor mutation burden compared to CM, and accordingly, lower expression of PD-L1 (Fig.
2a, b). These results indicated a relatively lower response to immune checkpoint inhibitors in MM compared to CM, as was suggested in a previous study [
53]. Interestingly, seven MM specimens and three MM specimens demonstrated higher tumor mutation burden or PD-L1 expression than CM, respectively (Fig.
2a, b, higher than median in CM cohort). The clinical relevance of high tumor mutation burden or PD-L1 expression in MM on immune checkpoint blockades still remains unknown, thus further investigation would reveal their potential as a predictor of response to immune checkpoint inhibitors.
Analyses on mutational spectrum and tumor mutation burden significantly differentiated primary anal MM from CM metastasized to the bowel. Although both melanoma types arise from melanocytes and grow in a similar mucosal microenvironment, the difference during their initiation, particularly the involvement of UV exposure, may lead to a distinct mutational landscape. Medical history or evidence of primary CM facilitates a definitive diagnosis between these types, however distinguishing these two types is occasionally challenging [
45]. Targeted NGS potentially facilitates a definitive diagnosis of anal melanoma, leading to relevant therapies for either CM or MM.
Acknowledgments
The authors thank Ms. Nousha Javanmardi for her editorial assistance and Dr. Sharon Huang, Mr. Nicolas Donovan and Mr. Garrett Cheung from Dept. of Translational Molecular Medicine, Division of Molecular Oncology (John Wayne Cancer Institute at Providence Saint John’s Health Center) for their kind advisory and technical assistance.