Germline variants in cancer susceptibility genes among patients with mucosal melanoma Authors: Afsaneh Amouzegar, Xiaogang Wu, James P Long, Gabriela W Chen, Justin W Wong, Sabitha Prabhakaran, Latasha Little, Curtis Gumbs, Neus Bota, Jared Malke, Julie Simon, Jianhua Zhang, Priyadharsini Nagarajan, Michael A Davies, Hussein Tawbi, Rodabe N Amaria, Isabella C Glitza Oliva, Alexandra P Ikeguchi, Shirley Su, Ehab Y Hanna, Roi Weiser, Brian Bednarski, Travis T Sims, Devarati Mitra, Jennifer Wargo, Jeffrey E Gershenwald, Paul Scheet, Yao Yu, Chad Huff, Kelly C Nelson, Jennifer L McQuade, P Andrew Futreal Published: 2026 ABSTRACT AbstractMucosal melanoma (MM) is a rare and aggressive melanoma subtype with poor outcomes and poorly understood risk factors, including the contribution of germline pathogenic variants (PVs). In this study, 346 MM patients treated at MD Anderson Cancer Center underwent germline sequencing of 322 known cancer susceptibility genes, with PV frequencies compared to population controls (gnomAD and TOPMed) using Fisher’s exact test. The cohort had a median age at diagnosis of 64 years and was predominantly female (62%) and White (84.0%). Anatomical subtypes of MM included gastrointestinal (36%), head and neck (33%), and genitourinary (31%). Among patients with somatic testing results (n = 303), KIT mutations (26%) were most common, followed by NRAS (17%), BRAFNon-V600 (6.6%), and BRAFV600 (2.6%). Germline PVs were identified in 37 patients (10.7%), most frequently in CHEK2 (n = 9, 23.0%), ATM (n = 8, 20.5%), and MITF (n = 6, 15.4%). MITF p.E318K (OR 6.0; 95% CI 2.2–13.2) and CHEK2 c.1100delC (OR 6.7; 95% CI 1.8–17.5) were significantly enriched compared to population control rates. Patients with germline PVs were more likely to have two or more affected first-degree relatives (49.0% vs. 29.1%, p = 0.019). These findings highlight a meaningful germline contribution to MM risk and support the incorporation of genetic testing in this population.Subject terms: Cancer, Genetics, Oncology Introduction Mucosal melanoma (MM) is the rarest form of melanoma, accounting for less than 1.5% of all melanoma cases in the United States1. Unlike cutaneous melanoma (CM), which arises from melanocytes in sun-exposed skin, MM originates from melanocytes located in mucosal membranes, most commonly in the head and neck (particularly the nasal and paranasal sinuses and oral cavity), anorectum, and female genital tract regions1,2. MM is characterized by distinct clinical features and often presents in more advanced stages at diagnosis. MM patients also often exhibit primary resistance to immune checkpoint inhibitors and have a significantly worse prognosis compared to CM patients, highlighting the need for further research into the pathogenesis of this disease3,4. From a molecular perspective, MM lacks the ultraviolet (UV)-associated mutational signature that characterizes CM, consistent with its development in sun-protected anatomical sites5. MM is characterized by a comparatively low tumor mutational burden (TMB) but a higher prevalence of chromosomal instability and structural variants, features that may contribute to its aggressive behavior and ability to evade immune surveillance5,6. Moreover, somatic mutations in BRAF are less common in MM, whereas activating mutations in KIT are more frequent6,7. Although the landscape of somatic mutations in CM and MM is well understood, our knowledge of germline pathogenic variants (PV) linked to hereditary melanoma and cancer-associated genetic syndromes continues to expand. Most publications addressing germline PVs and melanoma risk have reported on CM penetrance in largely White European or Australian cohorts. Early studies in this space focused on high melanoma penetrance genes such as CDKN2A/p16, p14/ARF, CDK4, and BAP1. With the identification of new moderate- to low-penetrance genes and expanded use of multigene panel testing, the landscape of melanoma PVs has expanded to include telomere maintenance genes (TERT, POT1), modifiers of cutaneous pigmentation (MC1R, MITF) and DNA repair genes (BRCA1/2, CHEK2, TP53, ATM). The reported risk of CM with the DNA repair gene PVs varies across studies but is generally low and may approach the risk of melanoma in the general population8,9. The overall prevalence of germline pathogenic variants in CM has been reported between 10-15% depending on whether the population is unselected or enriched for patients with significant personal or family history of melanoma10. Despite these advances in understanding germline PVs and CM risk, there has been minimal exploration of germline PVs in MM. Two small series exploring germline data for MM patients have suggested high prevalence of germline variants in this population11,12. One study, which specifically examined 16 MM patients with a significant family history of cancer, identified germline PVs in 50% of cases. The most frequently altered genes were CHEK2 and APC (each present in 12.5% of patients), with additional individual variants reported in MUTYH, ATM, RB1, and RECQL411. Defining the germline PV landscape for MM could potentially identify opportunities for personalized risk assessment, screening, early detection and personalized therapeutic strategies. To address this knowledge gap, we conducted a comprehensive analysis of PVs of known cancer susceptibility genes in a large, clinically annotated cohort of MM patients, who were not selected based on personal or family history of melanoma. Results Patient demographic and clinical characteristics A total of 346 patients with primary MM were identified. A summary of demographic and clinical characteristics of the cohort is shown in Table 1. The median age at diagnosis was 64 years (range 54–72), and the cohort was predominantly female (62%). Patients were predominantly White (84.0%), followed by Hispanic ethnicity (7.5%), Black (3.8%), and Asian (2.0%). The most frequently observed MM anatomical subtypes were gastrointestinal (36%), head and neck (33%), and genitourinary (31%). Clinical somatic mutation testing results were available for 303 patients (87.6%). Somatic mutations in KIT were detected in 26% of the cohort, followed by NRAS (17%), BRAFNon-V600 (6.6%) and BRAFV600 (E/K) (2.6%). 50% of tumors were triple wild type (TWT), i.e., lacking a somatic mutation in BRAF, NRAS and KIT. Table 1Summary of patient demographics, clinicopathologic features and family history in the total cohort and stratified by germline pathogenic variant statusGermline Pathogenic Variant StatusOverallN = 3461YesN = 371NoN = 3091p-value2Sex0.7Female214 (62%)24 (65%)190 (61%)Race> 0.9Asian7 (2.0%)1 (2.7%)6 (1.9%)Black13 (3.8%)1 (2.7%)12 (3.9%)Hispanic26 (7.5%)2 (5.4%)24 (7.8%)White289 (84%)32 (86%)257 (83%)Other11 (3.2%)1 (2.7%)10 (3.2%)Age64 (54, 72)62 (46, 70)64 (54, 72)0.3Primary tumor site0.5Gastrointestinal124 (36%)15 (41%)109 (35%)Genitourinary109 (31%)13 (35%)96 (31%)Head and Neck113 (33%)9 (24%)104 (34%)Melanoma In Situ component83 (24%)10 (27%)73 (24%)0.6Personal Hx of malignancy (Non-Melanoma)102 (29%)15 (41%)87 (28%)0.12Personal Hx of Other Melanoma18 (5.2%)2 (5.4%)16 (5.2%)> 0.91st Degree Relative with Malignancy0.2Y220 (64%)27 (73%)193 (63%)Unknown4 (1.1%)04 (1.3%)Number of 1st Degree Relatives with Malignancy0.0190122 (36%)10 (27%)112 (37%)1114 (33%)9 (24%)105 (34%)272 (21%)9 (24%)63 (21%)324 (7.0%)5 (14%)19 (6.2%)49 (2.6%)4 (11%)5 (1.6%)51 (0.3%)0 (0%)1 (0.3%)Unknown404Somatic mutationsKIT*76 (26%)9 (26%)67 (26%)> 0.9BRAFV600(E/K)**9 (3.0%)3 (8.8%)6 (2.2%)0.068Non-V600 BRAF†23 (7.7%)3 (8.8%)20 (7.5%)0.7NRASΨ49 (17%)7 (21%)42 (16%)0.5TWT143 (50%)14 (41%)129 (51%)0.31n (%); Median (IQR).2Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum test.*KIT mutations unknown in 51 (15%) of patients.**, † BRAF mutations unknown in 47 (14%) of patients.Ψ NRAS mutations unknown in 56 (16%) of patients.Hx history, TWT triple wild-type. Summary of patient demographics, clinicopathologic features and family history in the total cohort and stratified by germline pathogenic variant status 1n (%); Median (IQR). 2Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum test. *KIT mutations unknown in 51 (15%) of patients. **, † BRAF mutations unknown in 47 (14%) of patients. Ψ NRAS mutations unknown in 56 (16%) of patients. Hx history, TWT triple wild-type. Germline Pathogenic Variants We identified germline PVs in 37 out of 346 patients (10.7%). Most germline PVs (74.3%) were detected in genes involved in DNA damage response (DDR) pathway. Two patients had more than 1 germline PVs. Of the identified PVs most common were in CHEK2 (n = 9, 23.0%), ATM (n = 8, 20.5%), and MITF (n = 6, 15.4%), followed by BRCA2 (n = 2, 5.1%), and TP53 (n = 2, 5.1%) genes (Fig. 1, Table 2). The most frequently detected pathogenic variants were MITF (c.G952A:p.E318K) (n = 6, 15.4% of total pathogenic variants) at OR of 6.0 (95% CI 2.2–13.2), and CHEK2 (c.1100delC:p.T367fs) (n = 4, 10.2% of pathogenic variants) at OR of 6.7 (1.8–17.5) compared to population control rate (Table 3). Germline Pathogenic Variants Among Patients with Mucosal Melanoma. A Oncoplot of germline pathogenic variants and somatic mutations in melanoma patients. Rows represent individual genes with germline pathogenic variants (top panel), with the number of affected patients indicated in parentheses. Columns represent individual patients. Black squares denote the presence of a germline pathogenic variant in the specified gene. The lower panel displays selected somatic driver mutations (KIT, BRAF V600, non-V600 BRAF, and NRAS). Gray shading indicates missing somatic data. Clinical annotations are shown below, including primary tumor site (gastrointestinal, genitourinary, head and neck), sex (female, male), and age at diagnosis. B Stacked bar plots showing the proportion of patients with and without germline pathogenic variants across cancer history (personal and 1st degree family), somatic mutations, and disease site. PH: personal history; FH: family history. Figure created in R version 4.2.3 using the packages ComplexHeatmap (Fig. 1A) and ggplot2 (Fig. 1B). Table 2List of identified germline pathogenic variants in mucosal melanoma patient cohortGenePathogenic VariantNumber of patientsMITFp.E318K6CHEK2p.T367fs4p.T410fs2p.S400T1p.R188W1p.L464fs1ATMsplice1ATMp.L263fs1ATMp.R2034X1ATMp.R1875X1ATMp.L1327X1ATMp.A2067D1ATMp.M1L1ATMp.K468fs1ATRp.E1078X1ATRXp.D959fs1BRCA2p.A1689fs1BRCA2p.S1943fs1CHEK1p.W79X1ERCC2p.Y197fs1ERCC3p.R109X1ERCC4p.R589W1JAK1p.Q489X1NF1p.L1183R1PALB2p.V221X1POLEp.S1789fs1TP53p.R306X1TP53p.V73Wfs*501TSC1p.Q844X1XPO1p.Y4X1 List of identified germline pathogenic variants in mucosal melanoma patient cohort Table 3Frequencies of pathogenic MITF and CHEK2 variants in patients with mucosal melanoma compared to population allele frequencies based on gnomADGenegnomADMM Samplep-valueOROR 95% CI# CarriersTotal# CarriersTotalMITF p.E318K41211,401,48663460.000636.0(2.2,13.2)CHEK2 p.T367fs261149,29843460.003526.7(1.8,17.5)gnomAD Genome Aggregation Database, OR Odds Ratio. Frequencies of pathogenic MITF and CHEK2 variants in patients with mucosal melanoma compared to population allele frequencies based on gnomAD gnomAD Genome Aggregation Database, OR Odds Ratio. Association of germline PVs with clinical features and family history There were no differences in patient demographics or tumor clinicopathologic features between MM patients with and without germline PVs. Specifically, age did not differ significantly between germline PV-positive and PV-negative cohorts, including when stratified by DDR versus non-DDR genes (Supplementary Fig. 1). The frequencies of somatic KIT, NRAS, or BRAF mutations were similar between the two groups (Fig. 1A, B). Among the patients with an identified germline PV, 73% had a positive family history of malignancy in a first degree relative compared to 63% in patients without a germline PV. Among the germline PV positive cohort, 49.0% of patients had 2 or more affected first-degree family members compared to 29.1% in the germline negative cohort (p = 0.019). The most common non-cutaneous malignancies in first degree relatives in the germline positive cohort were breast, prostate, leukemia, colon and kidney cancer (Supplementary Table 2). Patients with a germline PV were more likely to report a personal history of other malignancy compared to those without a germline variant (41% vs. 28%), but this difference did not reach statistical significance (p = 0.12). Very few patients (n = 18, 5.2%) had a history of other melanomas (cutaneous or ocular melanoma), and this did not differ by germline PV status. Survival outcomes and treatment patterns The median overall survival (mOS) in the entire cohort was 4.9 years (95% CI 3.9 years–6.2 years, Fig. 2A) and was similar between the germline PV positive (6.6 years; 95% CI 3.9 years - N/A) and negative (4.8 years, 95% CI 3.7 years–5.9 years) patients (p = 0.81, Fig. 2B). In terms of systemic treatments, most patients were treated with immune checkpoint inhibitors (73%), followed by chemotherapy (50%) and targeted therapy (20%). There were no differences in treatment patterns between the germline PV positive and negative cohorts (Supplementary Table 3). Survival Analysis of Mucosal Melanoma Cohort by Pathogenic Variant Status. Overall survival in (A) the entire cohort, and (B) separated by germline pathogenic variant status. Figure created in R version 4.2.3 using the Survminer package. Discussion In this comprehensive analysis of 346 patients with MM, which to our knowledge is the largest retrospective cohort of MM patients characterized for germline PV to date, we report on the clinicopathologic and germline cancer susceptibility mutational landscape of this disease. Notably, the observed prevalence of 10.7% of germline pathogenic variants in our cohort exceeds the commonly cited 5% pre-test probability threshold that is frequently used to justify clinical genetic testing. This finding highlights the contribution of genetic factors to MM predisposition and has important implications for risk assessment and surveillance in MM patients. The characteristics of our cohort, including median age at diagnosis, female predominance, and ethnic distribution, were consistent with previous epidemiological studies and are representative of the MM patient population in the United States13. MM tumors were most commonly located in the gastrointestinal and genitourinary tracts, collectively accounting for 70% of the cases of the cohort, with the remaining 30% arising in the head and neck region. Somatic mutation profiling demonstrated KIT mutations in 22% of patients which is in line with the known enrichment of KIT alterations in MM11,14,15. Within our cohort of 346 MM patients, 10.7% of patients demonstrated a germline pathogenic variant. This aligns with a more limited cohort of 46 MM patients, which identified germline PVs in 13% of patients, including PVs in TP53, BRCA1, BRCA2, MUYTH, POT1, and CHEK212. The most frequently identified PVs in our cohort included CHEK2, ATM, and MITF, all of which have been implicated in hereditary cancer predisposition syndromes. However, their contribution to MM susceptibility have not been previously characterized. CHEK2 and ATM play key roles in DNA damage repair and are considered moderate-risk genes associated with increased risk for various cancers, including breast and pancreatic cancer16,17. Notably, the majority of germline alterations identified in our cohort affect genes involved in DNA damage response pathways. This pattern suggests that deficiencies in DNA repair may contribute to MM pathogenesis and raises the possibility that therapeutic strategies targeted to deficiencies in DNA damage and repair, such as PARP inhibitors, could have therapeutic potential in a subset of MM patients harboring these alterations. In our cohort, the most frequent specific germline PVs were MITF p.E318K and CHEK2 p.T367fs. These findings are particularly notable given that our cohort was not selected based on personal or family history of cancer. The MITF p.E318K is a gain-of-function variant that confers a ~ 4-fold increased risk of CM and has been implicated in renal cell carcinoma and pancreatic cancer18–20. Prior to our analysis, the MITF p.E318K variant had been reported only once in the context of MM in a patient with vulvar melanoma and a family history of cutaneous melanoma21. The CHEK2 p.T367fs (c.1100delC) truncating variant is seen in approximately 1-2% of the general population and is associated with a 2-fold increased risk of breast cancer in females22. The association between CHEK2 PVs and CM risk has been variable in the literature, with studies showing inconsistent or limited evidence of increased risk. While our study identified a high frequency of MITF and CHEK2 PVs, we did not observe other germline variants previously associated with hereditary CM, such as CDKN2A and BAP1, suggesting distinct germline susceptibility landscape for MM compared to CM. This study has several strengths, including the relatively large cohort size for a rare malignancy and a stringent assessment of deleterious germline variants. However, we do acknowledge limitations of this report. The retrospective design and absence of matched population controls limit our ability to infer causality or estimate precise penetrance of the identified variants. Further, given the targeted nature of the assessment, we cannot address the presence of novel germline susceptibility variants in genes not previously implicated in human cancer. Likewise, we do not address germline structural or non-coding variants in this analysis. Genome-wide association studies (GWAS), as well as whole genome germline sequencing in a prospective cohort are planned as a next step to comprehensively characterize the genetic architecture of MM. In summary, our findings demonstrate that germline PVs are present in patients with mucosal melanoma, with a prevalence exceeding thresholds commonly used to guide clinical genetic testing. These findings alongside emerging data from other recent reports, highlight the underrecognized role of inherited cancer susceptibility in MM and the importance of evaluating germline risk in this population. Although prospective validation is needed, our study provides a strong rationale for incorporation of germline testing into the multidisciplinary care of patients with MM, in particular patients with a strong family history of malignancy in first degree relatives. Methods Study population and data collection Patients diagnosed with MM between 1/1/1991 and 12/30/2023 with available banked peripheral blood mononuclear cells (PMBCs) for germline sequencing were identified through the institutional melanoma database at the University of Texas MD Anderson Cancer Center. The study was conducted in accordance with all relevant ethical regulations including the Declaration of Helsinki. The study was approved by the Institutional Review Board of the University of Texas MD Anderson Cancer Center under the following protocols: Melcore Melanoma Tissue Bank (Protocol No. LAB00-063); Longitudinal Biospecimen Acquisition for All Tumor Types and At-Risk Tissue (Protocol No. 2014-0938); Delineating germline, tumor and extrinsic factors driving mucosal melanoma risk and response (Protocol No. 2022-0428); and Patient Mosaic Data Acquisition, Aggregation, and Analysis (Protocol No. 2022-0404). Written informed consent was obtained from all participants prior to enrollment in each protocol. Cases of mucosal metastases from CM and melanomas of unknown primary were excluded. Clinical information extracted from patient charts included patient demographics such as sex, age, race, personal and family history of malignancy (in first-degree relatives); tumor characteristics, including primary anatomical site, results of targeted (clinical) somatic mutation testing for BRAF, NRAS, and KIT, and treatment history (immunotherapy, chemotherapy or targeted therapy). Germline sequencing and variant classification Germline sequencing was performed on genomic DNA derived from PBMCs using a next-generation sequencing panel of 322 genes (Supplementary Table 1) with established roles in hereditary cancer predisposition, as previously described23. Sequencing data were analyzed using a standardized bioinformatics pipeline to identify pathogenic and likely pathogenic variants. Pathogenic variants were determined by presence in ClinVar annotated as pathogenic/likely pathogenic, clearly deleterious (stop gain, frameshift, consensus splice variants), and/or previously reported and annotated in the literature as pathogenic. Statistical analysis Germline PV frequencies identified in our cohort were compared to population allele frequencies from the Trans-Omics for Precision Medicine (TOPMed)24 and The Genome Aggregation Database (gnomAD) using Fisher’s exact test. Odds ratios (OR) were computed in according to the following formula: Let pMM denote the probability of a germline PV (e.g., MITF p.E419K) in our mucosal melanoma cohort while pC is probability of germline PV in either TOPMed or gnomAD. Then the OR is pMM/(1-pMM) / (pC / (1-pC)). Patients were classified based on the presence or absence of a germline PV, and clinicodemographic features were compared between the groups using Pearson’s Chi-squared test and Wilcoxon rank sum test. P < 0.05 was considered statistically significant. Supplementary information Supplementary Information Acknowledgements This study received funding support from DOD grant W81XWH2210975, Melanoma Research Foundation, Andrew Sabin Family Foundation, Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the AIM at Melanoma Foundation, the NIH/NCI P50CA221703, the NIH/NCI P30CA016672, SPORE in Melanoma at University of Texas MD Anderson Cancer, the American Cancer Society and the Melanoma Research Alliance, Cancer Fighters of Houston, the Anne and John Mendelsohn Chair for Cancer Research, the Andrew M. McDougall Brain Metastasis Clinic and Research Program, philanthropic contributions to the MD Anderson Melanoma Moon Shots Program, the MD Anderson Patient Mosaic Program, Young Investigator Award from Melanoma Research Alliance (#570806), and National Cancer Institute and the National Center for Advancing Translational Sciences of the NIH (CCSG P30CA016672-46 and CCTS UM1TR004906). Author contributions Study concept and design: P.S., J.L.M., A.F.; Data collection and analysis: A.A., X.W., J.P.L., G.W.C., J.W.W., S.P., L.L., C.G., N.B., J.M., J.S., J.Z., P.N., S.S., Y.Y., C.H.; Writing - original draft: A.A., X.W., J.P.L., J.L.M., A.F.; Writing - review and editing: A.A., X.W., J.P.L., G.W.C., M.A.D., H.T., R.N.A., I.C.G., A.P.I., E.Y.H., R.W., B.B., T.T.S., D.M., J.W., J.E.G., P.S., K.C.N., J.L.M., A.F. Data availability All data generated or analyzed during this study are included in this published article and its supplementary information files. Competing interests Authors A.A., X.W., J.P.L., G.W.C., J.W.W., S.P., L.L., C.G., N.B., J.M., J.S., J.Z., P.N., Y.Y., C.H., H.T., R.N.A., I.C.G., A.P.I., S.S., E.Y.H., R.W., B.B., T.T.S., D.M., J.W., J.E.G., P.S., and K.C.N. declare no financial or non-financial competing interests. Author M.A.D. has been a consultant to Replimmune, Nurix, Roche/Genentech, Array, Pfizer, Novartis, BMS, GSK, Sanofi-Aventis, Vaccinex, Apexigen, Eisai, Iovance, Merck, and ABM Therapeutics. M.A.D. is also a scientific advisory board member for THERAtRAME, and he has been the PI of research grants to MD Anderson by Roche/Genentech, GSK, Sanofi-Aventis, Merck, Myriad, Oncothyreon, Pfizer, ABM Therapeutics, and LEAD Pharma. Author J.L.M. has received honoraria from BMS, Merck and Replimmune. Author A.F. is a scientific advisory board member for Scorpion Therapeutics. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Jennifer L. McQuade, Email: Dr.JMcQuade@gmail.com P. Andrew Futreal, Email: afutreal@mdanderson.org. Supplementary information The online version contains supplementary material available at 10.1038/s41525-026-00583-y. References Associated Data Supplementary Materials Supplementary Information Data Availability Statement All data generated or analyzed during this study are included in this published article and its supplementary information files.