Metastatic melanoma represents a major health burden in Australia and New Zealand. Relative to other cancers melanoma is highly-immunogenic and it provides a good target for immunotherapies. By exploring the immune infiltration of melanoma tumours using transcriptomic data, we aim to elucidate intercellular signalling and other molecular features which are associated with immune targeting and improved patient survival.
We have curated a natural killer (NK) cell gene signature from public data sources/tools including LM22 (CIBERSORT) and the LM7 (Tosolini et al.) gene sets. Using the gene-set scoring method singscore we infer the infiltration of NK cells within metastatic melanoma samples and examine how this varies relative to scores associated with other immune cell subsets. We contrast our results against transcriptomic data from the LM-MEL cell line panel to distinguish factors which may be produced by melanoma cells to modulate immune targeting.
Consistent with recent reports, we observe a strong association between NK cell and dendritic cell infiltration which is required for a robust T cell response. These scores show an intriguing association with epithelial and mesenchymal gene sets which capture melanoma phenotype switching, and with a TGF-β EMT signature which specifically captures mesenchymal behaviours induced by the immunosuppressive ligand TGF-β. Metastatic melanomas with evidence of mesenchymal-like behaviour induced by stimuli other than TGF-β, and robust NK cell infiltration, are associated with significantly improved patient survival. Exploring these data further, we identify a number of transcripts associated with reduced/improved immune targeting which may provide novel immunotherapy targets.
These results show that gene set scoring provides an intuitive approach to reduce the dimensionality of transcriptomic data sets in a manner which captures clinically-relevant phenotypic changes. Such approaches will be essential to leverage large public data sets which are becoming increasingly available and may help identify new targets for modulation with immunotherapy.