Germline pathogenic variants in a large convenience cohort of multiple melanoma subtypes.

Sarah E. Lochrin Memorial Sloan Kettering Cancer Center, New York, NY info_outline Sarah E. Lochrin, Yelena Kemel, Jessica Stuart, James William Smithy, Diana Mandelker, Parisa Momtaz, Walid Khaled Chatila, Nikolaus Schultz, Michael A. Postow, Zsofia Kinga Stadler, Alexander Noor Shoushtari

Authors Sarah E. Lochrin Memorial Sloan Kettering Cancer Center, New York, NY info_outline Sarah E. Lochrin, Yelena Kemel, Jessica Stuart, James William Smithy, Diana Mandelker, Parisa Momtaz, Walid Khaled Chatila, Nikolaus Schultz, Michael A. Postow, Zsofia Kinga Stadler, Alexander Noor Shoushtari Organizations Memorial Sloan Kettering Cancer Center, New York, NY Abstract Disclosures Research Funding No funding sources reported Background: A heritable component accounts for approximately 10% of melanomas. These estimates derive from populations deemed high risk for a germline pathogenic variant (gPV). The prevalence and clinical features of a broader melanoma patient population are unknown. Methods: Using the MSKCC IMPACT dataset, we identified a convenience cohort of all pts with melanoma who had undergone targeted tumor-normal sequencing (NCT01775072) for gPV in 76-90 genes. gPV were categorized by association with cancer predisposition and penetrance. Results: 701 pts with melanoma and germline sequencing demonstrated 105 pathogenic variants in 29 genes (Table). 14% of pts (n=99) harbored a gPV in a cancer susceptibility gene (gPV+); 49% high/moderately pathogenic, 32% low/recessive, and 19% uncertain. 6% of gPV+ pts had 2 gPVs. 77% of gPVs were involved in DNA damage repair pathways. The genes known to be associated with melanoma risk accounted for 9% of gPVs, inclusive of CDKN2A (2%), CDK 4 (1%), MITF (4%), and BRCA2 (2%). A further 15% were in genes with suggested, as yet undefined, susceptibility for melanoma; BRCA1 (6%), ATM (5%), BAP1 (2%) and PALB2 (2%). gPV+ frequency varied by melanoma subtype: unknown primary (19%, n=18), cutaneous (15%, n=64), mucosal (13%, n=13), acral (7%, n=2) and uveal (4%, n=2) (p=0.039). No significant difference in age, sex, race, stage at diagnosis, personal history of other cancer, or family history (FHx) of melanoma was seen between gPV+ and gPV-, on univariate analysis. 47% of gPV+ pts had either >1 primary melanoma, age<45 years or a FHx of melanoma (n=47), of which 14 pts had >1 risk factor. The presence of >1 melanoma and FHx of melanoma was associated with gPV+ (p=0.016). There was no significant difference in somatic TMB and MAPK pathway alterations between gPV+ and gPV-. In this cohort selected for high-risk melanomas, melanoma was the first presenting cancer for 83% of patients with gPV+. Assessment of biallelic loss to clarify if gPVs may be driving melanoma carcinogenesis is ongoing. Conclusions: This represents the largest cohort of sequenced matched germline and somatic analysis in melanoma. There is a 14% overall gPV prevalence that varies by melanoma subtype. The association of >1 primary melanoma, young age and a family history of melanoma with gPV+ is consistent with prior studies, however, would not capture 53% of gPV+ in this cohort. Melanoma represents the index cancer for the majority of gPVs, informing the need to expand germline testing in patients with melanoma. Pathway & Gene No. of gPVs (n=105) DNA Damage Sensing CHEK2 ATM NBN 16 5 3 Homologous Recombination Repair BRCA1 RAD51B-D RECQL (4) BRCA2 RAD50 PALB2 BAP1 BLM* MRE11 BRIP1 6 5 5 2 2 2 2 2 1 1 Nucleotide Excision Repair MUTYH* ERCC3 NTHL1* 11 7 2 Fanconi Anemia FANCC* FANCA* 4 2 Cell Cycle MITF CDK2NA CDK4 4 2 1 Cell Signaling APC I1307K NF1 SMARCB1 10 2 1 Citric Acid Cycle FH SDHC 3 1 *Monoallelic variants that only cause disease states if biallelic.

Clinical