Loss-of-function Variants In FSIP1 Identified By Targeted Sequencing Are Associated With One Particular Subtype Of Mucosal Melanoma

Background:Melanoma is a malignancy of pigment-producing cells called melanocytes,which can be found throughout the body（including in the skin,iris and rectum）.Mucosal melanoma accounts for approximately 1.3%of all melanomas and can arise from any mucosal tissue where melanocytes are present,although it most typically affects the mucosal epithelium of the respiratory,alimentary and genitourinary tracts.Due to its rarity,the mucosal melanoma subtype is poorly described,and its genetic characteristics are infrequently studied.The discovery or confirmation of new mucosal melanoma susceptibility genes will provide important insights for the study of its pathogenesis.Although mucosal melanoma is a special type of melanoma,its pathogenic mechanism is still inextricably linked with that of other melanoma types,and we designed this study accordingly.We performed deep targeted sequencing of 100 melanoma-related genes in39 mucosal melanoma samples followed by a gene-level enrichment analysis of the overall results.Materials and methods:A total of 39 formalin-fixed paraffin-embedded mucosal melanoma tissues were enrolled.The subtypes of mucosal melanoma selected in this study include the following 7 types:anorectal,esophageal,intestinal,sinonasal,vaginal,urinary tract and oral.All patients were self-reported Han Chinese.According to the Agilent website,the targeted sequencing probes was designed for 100 specific genes,and the DNA of 39 mucosal melanoma samples were extracted and sequenced by Illumina Hi Seq4000 platform.After preliminary data quality control,the BWA software was used to compare the sequencing sequence with the human reference genome sequence.SAMtools and Picard were used to sort the BAM files and perform duplicate marking,local realignment,and base quality recalibration to generate the final BAM file and calculate the sequence coverage and depth.SAMtools was used to SNVs and In Dels from single tumor samples.In addition to those removed by the default filters,SNVs and In Dels referenced in the 1000 Genomes Project and Ex AC with minor allele frequencies greater than 1%were removed.ANNOVAR is used for data annotation.The biallelic alleles were converted to plink format by Plink 1.9 for the subsequent analysis.Finally,the enrichment analysis of different groups was performed using SKAT toolkits.Results:In this study,we selected 100 genes and performed targeted sequencing of these genes at a coverage depth of at least 500×in the selected 39 mucosal melanoma samples.In the sequencing results,74 genes were found to have 484 LOF variants,including 6frameshift deletions in 6 genes,5 frameshift insertions in 5 genes,445 nonsynonymous SNVs in 73 genes,and 28 stop-gain mutations in 21 genes,averaged at least 1-6 samples per mutation.To understand whether these LOF mutation genes are associated with mucosal melanoma under different classifications,we conducted an enrichment analysis for every classification.We observed positive results based on the P value of the nominal statistical evidence（burden,SKAT,SKATO and chi tests,P<0.05）for all four detection methods in the enrichment analysis.Finally,the FSIP1 gene was found to be associated with oral mucosal melanoma(P_Chi=4.05×10^-2,P_burden=3.06×10^-2,P_skat=3.01×10^-2,P_skato=3.01×10^-2,OR=0.33,95%CI=0.15–0.72).However,beyond the significant associations mentioned above,no other associations were detected between the genes and the other mucosal melanoma subtypes（anorectal,esophageal,intestinal,sinonasal,vaginal and urinary tract）.Conclusions:We confirmed that FSIP1 is a susceptibility gene for oral mucosal melanoma.These findings contribute to a better genetic understanding of mucosal melanoma of different subtypes.The enrichment analyses described here provide insights into the importance of different genes in the pathogenesis of various subtypes of mucosal melanoma.