Genetic and Methylation Analysis of CTNNB1 in Benign and Malignant Melanocytic Lesions

Simple Summary Recurrent CTNNB1 exon 3 mutations have been recognized in the distinct group of melanocytic tumors showing deep penetrating nevus-like morphology and in 1–2% of advanced melanoma. We performed a detailed genetic analysis of difficult-to-classify nevi and melanomas with CTNNB1 mutations and found that benign tumors (nevi) show characteristic morphological, genetic and epigenetic traits, which distinguish them from other nevi and melanoma. Malignant CTNNB1-mutant tumors (melanoma) demonstrated a different genetic profile, grouping clearly with other non-CTNNB1 melanomas in methylation assays. To further evaluate the role of CTNNB1 mutations in melanoma, we assessed a large cohort of clinically sequenced melanomas, identifying 38 tumors with CTNNB1 exon 3 mutations, including recurrent S45 (n = 13, 34%), G34 (n = 5, 13%), and S27 (n = 5, 13%) mutations. Locations and histological subtype of CTNNB1-mutated melanoma varied; none were reported as showing deep penetrating nevus-like morphology. The most frequent concurrent activating mutations were BRAF V600 (55%) and NRAS Q61 (34%). Abstract Melanocytic neoplasms have been genetically characterized in detail during the last decade. Recurrent CTNNB1 exon 3 mutations have been recognized in the distinct group of melanocytic tumors showing deep penetrating nevus-like morphology. In addition, they have been identified in 1–2% of advanced melanoma. Performing a detailed genetic analysis of difficult-to-classify nevi and melanomas with CTNNB1 mutations, we found that benign tumors (nevi) show characteristic morphological, genetic and epigenetic traits, which distinguish them from other nevi and melanoma. Malignant CTNNB1-mutant tumors (melanomas) demonstrated a different genetic profile, instead grouping clearly with other non-CTNNB1 melanomas in methylation assays. To further evaluate the role of CTNNB1 mutations in melanoma, we assessed a large cohort of clinically sequenced melanomas, identifying 38 tumors with CTNNB1 exon 3 mutations, including recurrent S45 (n = 13, 34%), G34 (n = 5, 13%), and S27 (n = 5, 13%) mutations. Locations and histological subtype of CTNNB1-mutated melanoma varied; none were reported as showing deep penetrating nevus-like morphology. The most frequent concurrent activating mutations were BRAF V600 (n = 21, 55%) and NRAS Q61 (n = 13, 34%). In our cohort, four of seven (58%) and one of nine (11%) patients treated with targeted therapy (BRAF and MEK Inhibitors) or immune-checkpoint therapy, respectively, showed disease control (partial response or stable disease). In summary, CTNNB1 mutations are associated with a unique melanocytic tumor type in benign tumors (nevi), which can be applied in a diagnostic setting. In advanced disease, no clear characteristics distinguishing CTNNB1-mutant from other melanomas were observed; however, studies of larger, optimally prospective, cohorts are warranted.


Introduction
Studies over the past decade have yielded a more detailed understanding of the genetics of melanocytic tumors. Activating mutations of the MAPK pathway are present in most nevi and melanomas. Genetically, melanoma has been classified based on driver mutations activating the MAPK signaling pathway as (I) BRAF-mutant (50-60%), (II) RASmutant (20-30%), (III) NF1-mutant (10-15%) or (IV) triple (BRAF, NRAS and NF1) wild-type melanoma (~10%) [1,2]. Cutaneous melanoma harbors a larger number of mutations than any other major cancer entity as a result of UV-exposure. In both, benign and malignant melanocytic tumors, mutations in conjunction with MAPK mutations can occur. Another pathway playing a critical role in melanocyte biology is the Wnt/β-catenin signaling pathway. Inactivation of β-catenin (CTNNB1) in neural crest cells during embryogenesis can prevent development of the melanocytic lineage [3]. Activation of the pathway promotes both differentiation and expansion from neural crest progenitors to melanocytes [4]. Activating mutations of CTNNB1 have been reported in a range of cancers including melanoma [5]. Intracellular levels of CTNNB1 control canonical Wnt/β-catenin signaling. In a physiological setting, in the presence of Wnt ligands, CTNNB1 is translocated to the nucleus and activates the transcription of downstream target genes by binding with members of the Lef/Tcf transcription factors family [6,7]. In the absence of Wnt ligands, cytoplasmic CTNNB1 is targeted by a destruction complex that phosphorylates highly conserved serine/threonine residues located in exon 3 of CTNNB1, leading to degradation by the proteasome [7].
In melanocytic tumors, CTNNB1 mutations have been found to be present in almost all cases of deep penetrating nevi [8]. A recent review summarizing data from multiple next generation sequencing (NGS) approaches found only eight CTNNB1 mutations in 686 melanomas (1.2%). In this study, CTNNB1 mutations only occurred in BRAF or NRAS mutated melanomas, suggesting a cooperation between MAPK and Wnt/βcatenin signaling pathways [9]. Another large retrospective study including NGS data from 467 melanoma patients identified ten primary melanoma patients harboring a CTNNB1 mutation [7]. Here, concurrent CTNNB1 and MAPK mutations were found to not necessarily confer a deep penetrating nevi phenotype, and often progress to a metastatic stage [7]. The role of Wnt/β-catenin signaling in melanoma remains controversial. Although CTNNB1 has been reported to induce melanoma metastasis [10], it has also been described to limit invasion of melanoma in an experimental setting in both human and mice [11].
Deep penetrating nevi are rare, and even though most can be easily recognized by a well-trained pathologist, they can, on occasion, be mistaken for melanoma. In cases of deep penetrating melanocytic proliferations in which definitive diagnosis by morphology and immunohistochemistry (IHC) alone is difficult, molecular assays including screening for presence of CTNNB1 mutations as well as other alterations such as CDKN2A loss or TERT promoter mutations [12,13], may provide additional information to aid classification. Accurately classifying melanocytic tumors as benign or malignant has important implications for the patient in terms of prognosis, follow-up and treatment (including newly introduced adjuvant therapies, as targeted BRAF-inhibitors and MEK-inhibitors [14][15][16][17][18] and immune checkpoint therapies [19,20] which have shown great promise). However, Cancers 2022, 14, 4066 3 of 18 many melanomas still fail to respond to therapy or develop resistance to initially effective therapies [21,22]. A better understanding of which tumors will respond to therapy and how to identify and circumvent resistance in tumors would be of great clinical benefit to affected patients.
In this retrospective study, our aim was to explore to what extent CTNNB1 mutations are associated with certain clinical, histological, genetic, epigenetic and therapeutic features in melanocytic tumors. Mutation, copy number and methylation profiles were evaluated in CTNNB1-mutant melanocytic tumors and compared with those of CTNNB1-wild type tumors. In addition, the role of CTNNB1 mutations in a large cohort of advanced melanoma was analyzed to study potential associations with clinical characteristics, outcome and therapy responses [23].

Patient Selection
Patient medical history and data were retrieved from the medical databases/documentation system of the University Hospital Essen. The study was approved by and performed in accordance with the guidelines of the ethics committee of the University of Duisburg-Essen (BO-9589-20). Molecular testing was performed in patients included, with informed consent. To address all aims, three distinct cohorts were studied. (I) A cohort of seven CTNNB1-mutant melanocytic tumors was analyzed using genome-wide DNA methylation in conjunction with copy number variation (CNV) and mutation profiling (Sections 3.1 and 3.2). The control group consisted of eight benign nevi, eight malignant melanoma, and five Spitz nevi. Clinical characteristics of patients from the control group are shown in Supplementary Table S1. Additional information regarding these cases can be found in a previous manuscript [24]. The seven difficult to classify cases were either seen at our department or referred to our department from other institutions for further analysis. Tumors with deep-penetration nevus-like morphology characteristically show large cells with no maturation toward deeper tissue and an infiltrative growth pattern, expanding in an interstitial fashion into the tissue. These are traits that can also be seen in melanoma. In immunohistochemistry, deep penetrating nevus (DPN) like tumors can express HMB45 at certain amounts and demonstrate some level of reactivity to MIB (or Ki-67). All cases deemed not clearly benign by conventional histopathological analysis were passed on for genetic analysis.
(II) Clinical and mutational data from n = 38 CTNNB1-mutant melanoma patients were retrieved from routinely performed NGS melanoma panel analysis and the medical databases/documentation system of the University Hospital Essen (Sections 3.3 and 3.4). Only melanoma patients with metastatic disease (advanced melanoma) and therefore clinically confirmed malignant disease in stages IIIA or higher were included (see also Table 1). (III) Anti-PD1-treated melanoma patients were retrieved from [23]. Chi square and Fisher t tests were used for comparison of categorical variables as applicable. A Kruskal-Wallis test was used for continuous variables. Wilcoxon rank sum tests were used for continuous variables in R 4.2.0 (Section 3.5). For survival analysis, progression-free survival (PFS) and overall survival (OS) were defined as time from therapy start, until disease progression or death, respectively; if no such event occurred, the date of the last patient contact was used as endpoint of survival assessment (censored observation).

DNA Isolation and Targeted Sequencing
Sections of 10 µm-thick, were cut from formalin-fixed, paraffin embedded tumor tissues. The sections were deparaffinized and manually microdissected according to standard procedures. Genomic DNA was isolated using the Lotus DNA Library Prep Kit from IDT ® , according to the manufacturer's instructions. DNA capture-based, targeted next-generation sequencing was applied to analyze 611 target genes known to be mutated in cutaneous melanoma and other human cancers (see Supplementary Table S2

Mutation Sequence Analysis
CLC Cancer Research Workbench from QIAGEN ® was used for sequence analysis, as has been previously reported [25][26][27]. In brief, the analysis workflow described included adapter trimming and read pair merging, before mapping to the human reference genome (hg19). Detection of insertions and deletions as well as single nucleotide variants followed. Additional information regarding potential mutation type, known single nucleotide polymorphisms and conservation scores was obtained by cross-referencing various databases (COSMIC, ClinVar, dbSNP, 1000 Genomes Project, HAPMAP, and PhastCons-Conservation_scores_hg19). Further analysis of csv files was performed by applying R (version 4.0.3). The mean coverage of the targeted sequencing region achieved in targeted DNA sequencing of all CTNNB1 mutant samples 1776.3 reads with 99.9% of the target region sequenced with a coverage ≥ 30 reads. Mutations were considered if coverage of the mutation site was ≥30 reads, ≥10 reads reported the mutated variant and the frequency of mutated reads was ≥10%. Copy number variations determined by targeted sequencing were detected with CLC Cancer Research Workbench (QIAGEN ® ) and are based on the following algorithms [28][29][30]. A ≥1.7 absolute fold copy number change involving a region with greater than 30 target sequences was chosen as a cut-off for detecting copy number variations.

DNA-Methylation Profiling and Copy Number Analysis
Array-based copy number and methylation analysis required 500 ng of isolated DNA and was performed on (Spitz nevi (n = 5), benign nevi (n = 8), malignant melanoma (n = 8), and CTNNB1 mutant melanocytic tumors (n = 7). The HumanMethylationEPIC (EPIC) bead-based microarrays from Illumina were used to obtain genome-wide methylation data, according to the manufacturer's instructions [31,32]. Methylation analysis using EPIC arrays was performed by the Genomics and Proteomics Core Facility, Heidelberg, Germany. Unnormalized DNA methylation data were obtained as IDAT files, which were used as input to the RnBeads software package implementing a comprehensive workflow for quality control, preprocessing and analysis of data from DNA methylation microarrays [32,33]. In brief, DNA-methylation data were normalized by performing background correction and dye bias correction, whereupon low-quality and potentially biased measurements, e.g., from probes obtained with too few microarray beads, probes with low signal/noise ratio (detection p-value), probes containing single nucleotide polymorphisms and cross-reactive probes, are removed or masked. 10,000 sites, most variable across all samples were used for both, principal component analysis and clustering analysis and visualized as heatmaps [32,34]. The copy number profile was generated from the array data using the "conumee" R package in Bioconductor (http://www.bioconductor.org/packages/release/bioc/html/conumee.html (package version 1.26.0), accessed on 20 June 2022). The conumee median signals per bin were summarized in chromosome arms, and the gains and losses were called for arms with summarized median signal above 0.1 or below −0.1, respectively.

Reference-Free Methylome Deconvolution Using MeDeCom
DNA methylation data of the bulk tumor samples of the cohort were investigated using the reference-free MeDeCom algorithm that dissects DNA methylation data into major components of variation, called latent methylation components (LMC) [35]. DNA methylation data of patients were processed according to a recently published protocol [36]. The protocol selected the 20,000 most variably methylated CpG sites across the samples as input to MeDeCom. LMCs were functionally annotated within the FactorViz platform [36].

CTNNB1 Mutations in Difficult-to-Classify Benign and Malignant Melanocytic Tumors
Challenging melanocytic cases sent for reference histopathology from experienced dermatohistopathologists were sequenced using NGS. Seven cases with CTNNB1 mutations were identified, two classified as malignant, one as a "deep blue nevus-like melanoma" (case 3) and a "malignant melanoma morphologically resembling a deep penetrating blue nevus" (case 6). The other cases had been classified as a mainly intradermal Spitz nevus (case 1), a deep penetrating blue nevus (case 2), a combined nevus (case 4), a deeppenetrating nevus (case 5), and a congenital nevus cell nevus (case 7) ( Table 1). Six patients were male and two tumors were localized on the head. Two tumors harbored additional BRAF V600E mutations and one showed an NRAS Q61R mutation. Both melanomas showed mutations in the TERT promoter region, and one additionally harbored an NF1 mutation (Table 1). (Full clinical details of the two melanoma cases are included in the Supplementary Material).

Methylation Profiling with Comprehensive Copy Number Analysis
Methylation analysis was performed on the seven described CTNNB1-mutant tumors and the profiles were compared to control groups consisting of eight benign nevi, eight malignant melanomas and five Spitz nevi [24] ( Figure 1A). HumanMethylationEPIC data were preprocessed using RnBeads workflow to obtain genome-wide DNA methylation profiles [32]. To assess the global trends of DNA methylation variability of CTNNB1-mutant Cancers 2022, 14, 4066 6 of 18 melanocytic tumors, we applied principal components analysis (PCA). We observed a distinction between the histopathological entities (nevi, Spitz nevi, and melanoma) within the first two principal components ( Figure 1B). CTNNB1-mutant melanocytic tumors showed a methylation profile clustering between Spitz and normal nevi. The two malignant CTNNB1mutant tumors (ID 3 + 6) clustered towards the melanoma group. Hierarchical clustering analysis of average methylation profiles of 10,000 gene regions confirmed two distinct groups: (1) benign CTNNB1-mutant melanocytic tumors, nevi, Spitz nevi, and (2) malignant CTNNB1-mutant tumors and CTNNB1-wild type melanoma ( Figure 1D). A clear separation was also observed regarding CNV, which were not detected in benign CTNNB1mutant tumors, but were present in CTNNB1-mutant melanoma ( Figure 1B, Table 1). To better understand cellular composition of CTNNB1 mutant samples and control groups, bulk DNA methylation data were deconvoluted assessed using reference-free algorithm MeDeCom that dissects DNA methylomes into major components of variation, called latent methylation components (LMC), and estimates their relative proportions. The crossvalidation error pointed at five LMCs (parameter k) during MeDeCom parameter selection and the regularization parameter λ value of 0.01 (Supplementary Figure S1A,B). The proportions of the five LMCs (LMC1-5) in all samples are visualized in Figure 1E. Hierarchical clustering analysis of LMC proportions revealed well-separated clusters corresponding to the benign and malignant tumors ( Figure 1E). ( Figure 1D). A clear separation was also observed regarding CNV, which were not detected in benign CTNNB1-mutant tumors, but were present in CTNNB1-mutant melanoma ( Figure 1B, Table 1). To better understand cellular composition of CTNNB1 mutant samples and control groups, bulk DNA methylation data were deconvoluted assessed using reference-free algorithm MeDeCom that dissects DNA methylomes into major components of variation, called latent methylation components (LMC), and estimates their relative proportions. The cross-validation error pointed at five LMCs (parameter k) during MeDeCom parameter selection and the regularization parameter λ value of 0.01 (Supplementary Figure S1A,B). The proportions of the five LMCs (LMC1-5) in all samples are visualized in Figure 1E. Hierarchical clustering analysis of LMC proportions revealed well-separated clusters corresponding to the benign and malignant tumors ( Figure 1E).

Clinical Characteristics of CTNNB1 Mutated Melanoma Patients
Data acquired from routinely performed NGS panel analysis of histopathologically clearly diagnosed melanomas analyzed between 2014 and 2021 were assessed to identify samples harboring CTNNB1 mutations. Mutations were found to be distributed across the gene, however recurrent mutations were found in the known N-terminal hotspot region

Clinical Characteristics of CTNNB1 Mutated Melanoma Patients
Data acquired from routinely performed NGS panel analysis of histopathologically clearly diagnosed melanomas analyzed between 2014 and 2021 were assessed to identify samples harboring CTNNB1 mutations. Mutations were found to be distributed across the gene, however recurrent mutations were found in the known N-terminal hotspot region on exon 3 ranging from amino acid 25 to 46 (Figure 2A,B) [5,37]. The remainder of the identified mutations were distributed randomly across the gene and are assumed to be passenger mutations (Figure 2A,B). In total, 38 mutations in melanoma from 38 patients in the exon 3 hotspot of CTNNB1 were identified. The majority of patients (n = 27, 71%) were male, and the median age at melanoma diagnosis was 59 years (range 39-90 years). Eighteen percent (n = 7) of patients had a nodular melanoma, 13% (n = 5) each a superficial spreading melanoma or a melanoma of unknown primary (MUP), 8% (n = 3) an acrolentiginous melanoma (ALM), 3% (n = 1) had a spitzoid melanoma, and 45% (n = 17) had an unspecified histopathological subtype ( Table 2). Forty-eight percent (n = 18) of primary melanoma were ulcerated. Primary melanomas were localized on the lower extremity in 37% (n = 14), trunk (32%, n = 12) and the head/neck region (13%, n = 5) ( Table 2). Forty-two percent of patients (n = 16) were stage IIIA or higher at diagnosis. During the disease course, 66% (n = 25) of patients with a CTNNB1 mutation developed lymph node metastasis, 32% (n = 12) lung metastasis, 29% (n = 11) hepatic metastasis, 29% (n = 11) metastasis within the central nervous system (CNS), and 11% (n = 4) bone metastasis. Nine patients (24%) each received either targeted therapy (TT) or immune checkpoint inhibitor-based therapy as a first-line treatment. Patients treated with TT showed partial response (PR, n = 2), stable disease (SD, n = 2) and progressive disease (PD, n = 3). In two cases the response was unknown. Patients treated with immune checkpoint inhibitor-based therapy showed PR (n = 1) and PD (n = 8). Other therapies administered for advanced disease included chemotherapy (n = 1, PD), the NIPAWILMA trial (n = 1, PD) [38], and the TriN 2755 trial (n = 1, SD) [39].

CTNNB1 Mutation Status and Transcriptomic Alterations in an Anti-PD1 Monotherapy Treated Melanoma Cohort
A previously described cohort of 144 melanoma patients treated with anti-PD1 monotherapy and mutational and transcriptomic data were used to investigate therapy response and gene expression profiles in CTNNB1 mutant patients (23). Nine (6.2%) showed a mutation in exon 3 of CTNNB1, of which transcriptomic data were available in eight cases. No significant differences within clinical characteristics and disease course were found between CTNNB1 mutant and non-mutant patients (Table 4). Therapy response to anti-PD1 monotherapy showed an overall response rate (ORR) of 67% (6/9 patients) in patients with CTNNB1-mutant melanoma compared to 36% (49/135 patients) in those with CTNNB1-wild type tumors. Median number of total mutations, nonsynonymous mutations, clonal and subclonal mutations were higher in CTNNB1 mutant melanoma patients, albeit not significantly (Table 4). Differentially regulated genes between CTNNB1 mutant and non-mutant melanoma patients can be found in Supplementary Table S3. As expected, Enrichr analysis (https://maayanlab.cloud/Enrichr/ (accessed in February 2022)) of the 100 top differentially expressed genes revealed the Wnt-β Catenin Signaling pathway as the second pathway after fatty acid metabolism in the MSigDB Hallmark 2020 representing well-defined biological states or processes (Supplementary Table S4). In addition, the Wnt signaling pathway and pluripotency was the second pathway in BioPlanet 2019 and WikiPathway 2021 Human. Comparison of transcriptomic expression (measured in transcripts per million [TPM]) of multiple genes involved in this Hallmark WNT β-catenin signaling pathway (Supplementary Table S5) showed significant differences in the expression of CTNNB1-mutant and CTNNB1-wild type melanoma patients ( Figure 3A). Expression of AXIN2 (p ≤ 0.0001), NKD1 (p = 0.003), TP53 (p = 0.003), HEY1 (p = 0.03), PSEN2 (p = 0.001), and CUL1 (p = 0.01) was significantly higher in CTNNB1 mutant melanoma, supporting these mutations leading to an over activation of the Wnt pathway. Transcriptomic expression of CTNNB1 was not elevated in CTNNB1-mutant melanoma. However, generally in the entire cohort, higher CTNNB1 expression levels were correlated with expression of AXIN1 (p = 0.009), AXIN2 (p ≤ 0.001), HEY1 (p ≤ 0.0001), PSEN2 (p ≤ 0.0001), PPARD (p ≤ 0.0004), and CUL1 (0.04) ( Figure 3B). No significant difference in OS and PFS between CTNNB1-mutant and CTNNB1-wild type patients was observed ( Figure 3C,D).  0.001), and CUL1 (p = 0.01) was significantly higher in CTNNB1 mutant melanoma, supporting these mutations leading to an over activation of the Wnt pathway. Transcriptomic expression of CTNNB1 was not elevated in CTNNB1-mutant melanoma. However, generally in the entire cohort, higher CTNNB1 expression levels were correlated with expression of AXIN1 (p = 0.009), AXIN2 (p ≤ 0.001), HEY1 (p ≤ 0.0001), PSEN2 (p ≤ 0.0001), PPARD (p ≤ 0.0004), and CUL1 (0.04) ( Figure 3B). No significant difference in OS and PFS between CTNNB1-mutant and CTNNB1-wild type patients was observed ( Figure 3C,D).

Discussion
CTNNB1 mutations occur in both benign and malignant melanocytic tumors with a deep penetrating nevus-like phenotype. We identified CTNNB1 mutations in two types of melanocytic tumors, benign nevi and malignant melanoma. Detailed histological, mutation, copy number and methylation analysis can clearly distinguish benign from malignant tumors. In addition, we investigated recurrent CTNNB1 exon 3 mutations in the largest cohort reported to date to determine whether these mutations are associated with specific features relevant in clinical patient management.
Mutations of the β-catenin pathway have been reported to transform the phenotype of a BRAF-mutated common nevus into that of a deep penetrating nevus, including increased pigmentation, cell volume, and cyclin D1 levels in the nucleus [8]. Mutational activation of the MAP kinase and β-catenin pathways are practically pathognomic of the characteristic DPN phenotype. Data have also suggested that constitutive β-catenin pathway activation promotes tumorigenesis by overriding dependencies on the microenvironment that constrain proliferation of common nevi, with DPN-like melanoma harboring additional oncogenic mutations; further, these data identified DPN as an intermediate melanocytic neoplasm, positioned between benign nevus and DPN-like melanoma [8]. Our histopathologically challenging cases confirmed that in all seven cases, additional mutations in either BRAF, NRAS, NF1 or further MAP kinase related genes were present. Interestingly, mutations in the TERT promoter region were only present in tumors identified as melanoma, and not in benign CTNNB1-mutant melanocytic tumors. Methylation profiling allowed a clear differentiation between benign and malignant (ID 3 + 6) CTNNB1-mutant tumors, underlining the potential of molecular and methylation analysis for further characterization of challenging cases.
Histopathologic evaluation remains the gold standard to classify melanocytic tumors and assess their likely clinical/biological potential. In most cases, including deep penetrating nevi, conventional histologic analysis is sufficient to distinguish benign from malignant tumors. However, in some histologically ambiguous tumors with deep penetrating morphology pathologic classification and determination of biological potential may not be clear-cut. In these difficult cases, genetic analysis may be a helpful additional tool in classifying deep penetrating tumors, as mutation profiles differ between primary melanomas and benign melanocytic tumors. The presented cases illustrate the potential diagnostic value of mutation profiling in a clinical setting.
Activating mutations (i.e., BRAF, NRAS, etc.) are found in both benign and malignant tumors, i.e., nevi and melanoma. A common theory is the acquisition of additional genetic events lead to tumors progressing, eventually tipping the balance from benign to malignant proliferations. Other potentially relevant events, such as DNA replication errors, have been discussed in odontogenic cysts and tumors which are also mainly benign despite harboring activating MAP Kinase or CTNNB1 mutations [42].
To assess the role of CTNNB1 mutations in advanced melanoma, we screened our large genetic melanoma database identifying 38 tumors-to our knowledge the largest cohort of CTNNB1-mutated advanced melanoma reported to date. As described previously, these tumors are rare [7,9]. Oulès et al., reported three NMM (30%), three SSM (30%), two lentigo malignant melanoma (LMM) (20%), one ALM (10%) and one deep-penetrating nevus-like melanoma (10%). Our cohort included 18% NMM, 13% SSM, 13% MUP, and 8% ALM, demonstrating a comparable distribution. In many cases, a specific melanoma subtype was not reported. However, our data and previous studies underline that melanocytic tumors harboring CTNNB1 mutations often do not have a deep penetrating phenotype.
Comparing CTNNB1 mutation frequencies with overall low mutation numbers is difficult, but the distribution we observed is comparable to previous reports. The most frequent CTNNB1 mutations we observed were in S45, G34, T41 and S33 (found in 34%, 13%, 11% and 5% in our and 60%, 10%, 20% and 10% of cases in the Oulès et al., cohort, respectively) [7]. Concurrent mutations present were similar to results by Oulès et al., where an additional BRAF mutation was present in 55% of patients in our cohort (compared to 60% in [7]), but NRAS mutations were more frequent in our patients (34% compared to 20% in [7]). In our cohort, 4/38 patients had no additional mutations in BRAF or NRAS genes, showing that [9], CTNNB1 mutations can occur in tumors not harboring these gene mutations [7]. Two of these four melanomas had mutations in NF1, one in KIT and one in GNA11 (data not shown). A high rate of co-occurrence of MAPK-activating mutations (BRAF/NRAS/NF1) and CTNNB1 mutations favors the hypothesis that mutations in CTNNB1 display Wnt/β-catenin signaling proliferative hallmarks and cooperate with MAPK pathways [9].
For patients with advanced melanoma in whom CTNNB1 mutations are unexpectedly discovered during routine molecular profiling, the extent to which they might impact patient therapy may be the most important question to consider. We found better therapeutic responses in patients receiving targeted therapy regimes (ORR 22%; DCR 44%, PR 22%) compared to immune checkpoint inhibition (ORR 11%; DCR 11%, PR 11%). These data are consistent with the previously described synergistic activity between Wnt/β-catenin signaling activation and BRAF inhibitors to reduce melanoma growth in vitro and in vivo [43]. Oncogenic signals have been postulated to mediate cancer immune evasion and resistance to immunotherapies [44]. Data has suggested active β-catenin signaling results in T cell exclusion and lack of T cell infiltrate driving resistance to anti-PD-L1 and anti-CTLA-4 immunotherapies [45]. Within this work, Spranger et al. identified the Wnt/β-catenin pathway as the first defined tumor-intrinsic oncogene pathway that can abort the induction of antitumor T cell responses, prevent the T cell-inflamed tumor microenvironment, and generate resistance to checkpoint blockade therapy [46]. Using transcriptomic data from a cohort of >700 melanoma patients (primaries and metastasis), Nsengimana et al. could show that low-immune/β-catenin high expressing tumor patients show fewer pathologist-reported brisk tumor infiltrating lymphocytes (TILs) and significantly worsened melanoma-specific survival, underlining oncogenic potential of the Wnt pathway [47]. Significant changes on the transcriptomic level of genes involved in the Wnt/β-catenin signaling pathway underline a biological regulation of this pathway in CTNNB1-mutant melanoma. The ORR of 11% in our cohort would support CTNNB1-mutant tumors responding poorly to immune checkpoint therapy. However, performing an additional analysis of outcomes of CTNNB1mutant and CTNNB1-wild type melanoma patients from a recently published anti-PD1 monotherapy-treated melanoma cohort [23] an ORR of 67% (DCR 78%) in CTNNB1-mutant compared to 36% (DCR 50%) in CTNNB1-wild type melanoma patients, was observed, arguing immune checkpoint inhibition therapy can be effective for patients with CTNNB1mutant tumors. However, as the number of treated patients in both our and the Liu et al. study remain limited, larger studies are required.
Based on our and existing data to date, we believe no clear-cut recommendation concerning therapeutic approach or prognosis concerning survival can be made for CTNNB1mutant melanoma. Both types of therapy, targeted and immune-checkpoint inhibition, have shown efficacy. Larger studies, optimally in a prospective fashion will be required to further elucidate if CTNNB1 mutation status should be clearly linked to specific systemic therapy recommendations in advanced melanoma patients. A limitation of the study is the low number of patients. Studies with larger numbers of difficult to classify melanocytic lesions as well as CTNNB1 mutated melanomas may offer further insights. Considering performing both NGS sequencing and methylation arrays is cost intensive and not universally available, larger cohort studies may also help identify a selection of relevant gene mutation and methylation sites enabling a more focused cost-effective analysis.
In summary, we report the largest cohort of CTNNB1-mutated melanocytic tumors identified so far. CTNNB1 mutations can be found in difficult-to-classify tumors with deep penetrating morphology and additional molecular and methylation analysis can help differentiate between benign and malignant tumors in these cases. CTNNB1-mutant melanomas were found to originate from different locations and only rarely demonstrated a deep penetrating phenotype. Therapeutic responses to both targeted and ICI therapy were observed.

Conclusions
-Mutation analysis in conjunction with methylation analysis can be a diagnostic aid in determining the dignity in some cases of deep penetrating melanocytic tumors -CTNNB1-mutant melanoma comprises~1-2% of melanoma -Histologic characteristics can show a deep penetrating nevus, but can also be any melanoma subtype Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cancers14174066/s1. Figure S1. Parameter selection for MeDeCom analysis in the total cohort. A. Cross validation (CV) error plotted against the number of LMCs k. Cross validation error tends to decrease with more components being included and we selected k = 5 as the value where the error starts to level out. B. Selection of the regularization parameter λ for k = 5. We selected λ = 0.01 as the point where the cross-validation error is still low, while the objective function tends to increase. ID number of patients depicted. Lambda values of 0.01 were used. LMC, latent methylation components. Figure S2. An extensive melanocytic tumor consisting of two components is seen. The covering epithelium is inconspicuous and shows no melanocytes. Underlying and reaching into the subcutaneous adipose tissue there are melanocytes with slightly increased pigment content aggregated in large nests, associated with interspersed melanophages. There is also a larger population of very strongly pigment-bearing partly spindle cell, partly large epithelioid cell melanocytes in large nests and strands. In the nevoid part, occasional mitoses are seen (A, measuring bar represents 4 mm). Immunohistochemical staining for Ki67 (B), tumor markers S100 (C) MelanA (D), and HMB-45 (E), as well as BCAT (F), and PHH3 (G) were performed. Table S1. Clinical characteristics of patients from the control group. Table S2A. Targeted next generation sequencing panel. Table S2B. 611 oncogene Panel Genes. Table S3. Differentially regulated genes