The Neoantigen Landscape of Mycosis Fungoides.

Background: Mycosis fungoides (MF) is the most common cutaneous T-cell lymphoma, for which there is no cure. Immune checkpoint inhibitors have been tried in MF but the results have been inconsistent. To gain insight into the immunogenicity of MF we characterized the neoantigen landscape of this lymphoma, focusing on the known predictors of responses to immunotherapy: the quantity, HLA-binding strength and subclonality of neoantigens. Methods: Whole exome and whole transcriptome sequences were obtained from 24 MF samples (16 plaques, 8 tumors) from 13 patients. Bioinformatic pipelines (Mutect2, OptiType, MuPeXi) were used for mutation calling, HLA typing, and neoantigen prediction. PhyloWGS was used to subdivide malignant cells into stem and clades, to which neoantigens were matched to determine their clonality. Results: MF has a high mutational load (median 3,217 non synonymous mutations), resulting in a significant number of total neoantigens (median 1,309 per sample) and high-affinity neoantigens (median 328). In stage I disease most neoantigens were clonal but with stage progression, 75% of lesions had >50% subclonal antigens and 53% lesions had CSiN scores <1. There was very little overlap in neoantigens across patients or between different lesions on the same patient, indicating a high degree of heterogeneity. Conclusions: The neoantigen landscape of MF is characterized by high neoantigen load and significant subclonality which could indicate potential challenges for immunotherapy in patients with advanced-stage disease.