Abstract
Introduction: The most common subtype of cutaneous T-cell lymphoma (CTCL), mycosis fungoides (MF), is characterized by proliferation of malignant T-cells in the skin. In the more clinically aggressive folliculotropic variant of MF (FMF), malignant T-cells localize to follicular epithelium while in classic MF (CMF), they infiltrate the dermis and epidermis (Gerami et al., Arch Dermatology 2008; Agar et al., J Clin Oncol 2010). How the localization of neoplastic T cells to the follicular niche contributes to the clinical aggression in FMF is unclear.
Methods: To uncover the tumor microenvironmental differences between perifollicular FMF and dermal CMF regions, we analyzed 9 mycosis fungoides lesional skin biopsies (3 FMF, 6 CMF) using spatial transcriptomics. We performed and analyzed transcriptomic data from 71 regions of interest (ROIs) in FMF patients to 123 ROIs in CMF patients obtained from all cells using the Nanostring GeoMx Digital Spatial Profiler. Additionally, we compared data obtained from CD4+ cells, which include the predominant malignant T-cells, and CD68+ cells, which represent macrophages in the tumor microenvironment. The FMF ROIs were taken from perifollicular locations, while CMF ROIs were from the dermis. Our bioinformatics analyses included differential gene expression analysis, spatial deconvolution, gene set enrichment analysis (GSEA), and single sample GSEA (ssGSEA).
Results: Our ssGSEA analysis revealed that FMF CD4+ T cells showed higher activity in metabolic pathways (glycoprotein metabolic process – Gene Ontology, adjusted p = 0.044; response to EIF2AK4 GCN2 to amino acid deficiency - Reactome, adjusted p = 0.030), cholesterol transport pathways (positive regulation of cholesterol efflux – Gene Ontology, adjusted p = 0.030), and nutrient deprivation response pathways (cellular response to starvation - Reactome, adjusted p = 0.041) compared to those in CMF. Additionally, in GSEA and differential expression analysis, perifollicular cells in FMF exhibited increased activity in pathways and genes related to antigen presentation (antigen processing and presentation – Gene Ontology, NES = 4.34, p = 0; antigen processing and presentation of peptide or polysaccharide antigen via MHC class II – Gene Ontology, NES = 3.74, p = 0; antigen processing and presentation of peptide antigen via MHC class I – Gene Ontology, NES = 2.70, p = 0.0009) and inflammatory signaling (SPP1, log2FC = 1.95, p <0.0001; RASGEF1B, log2FC = 0.83, p <0.0001; NCF2, log2FC = 0.85, p = 0.013; neutrophil degranulation – Reactome, NES = 4.47, adjusted p = 0; cytokine production – Gene Ontology, NES = 3.15, adjusted p = 0). However, macrophages in FMF had a less inflammatory phenotype with decreased IL-2 (Reactome, NES = -1.84, adjusted p = 0.046) and interferon signaling (Reactome, NES = -2.02, adjusted p = 0.020) and decreased expression of genes such as KIR3DL1, CCL8, and IL21R (log2FC = -1.82, -1.16, -1.14 and p = 0.007, 0.024, and 0.028 respectively). These findings are further underscored by the results of spatial deconvolution, which showed a notable increase in the abundance of myeloid dendritic cells (mDCs) in FMF compared to CMF (p = 0.012), and a decrease in naive B cells (p = 0.031) and CD4+ memory T cells (p = 0.0005).
Conclusions: Our analyses revealed that, compared to their dermal counterparts in classical MF, malignant CD4 T cells in the follicular microenvironment of FMF exhibit survival advantages stemming from their elevated metabolic and inflammatory activity and a heightened response to starvation. On the other hand, the increased presence of tumor-supportive mDCs, and macrophages with an anti-inflammatory phenotype in follicular niche reveals a compromised antitumor immune response that may facilitate immune evasion in FMF. These observations provide insight into the mechanisms underlying the more aggressive clinical features of FMF versus CMF.
