Immune gene signatures and integrative spatially-resolved digital profiling of prognostic biomarkers for mesothelioma.

Abstract

e20071 Background: Malignant mesothelioma has been an incurable disease with few effective therapies. While PD-1 targeted therapies have elicited some patient responses, the overall response rate for mesothelioma is low. Since mesothelioma is derived from the mesothelium of the lung, we hypothesize that immune cells in the tumor microenvironment (TME) may behave differently than other solid tumors that are responsive to immunotherapy. Here we characterize prognostic immune gene signatures and spatial protein expression in the mesothelioma TME. Methods: 50 FFPE mesothelioma samples were analyzed using the NanoString PanCancer IO360 assay which measures expression of 770 genes, including the abundance of 14 immune cell types and of 32 IO signatures. All genes and signatures were correlated with overall survival (OS). GeoMx digital spatial profiling (DSP) was performed on 40 samples assessing the protein expression along 12 geometric circular regions-of-interest (ROI). Tissue slides were stained with a combination of fluorescent-labeled antibodies (pan-cytokeratin, CD3, CD68) and a panel of 38-antibodies each conjugated to a unique UV-photocleavable DNA barcode. UV light was applied to the defined ROI, which releases the DNA barcodes for quantitation on the nCounter platform. Results: Unsupervised clustering of samples based on gene signatures showed two distinct groups; one, with low expression of lymphocyte activation/exhaustion signatures and the second, with moderate expression of immune signatures. Two samples had high expression of all immune gene signatures, also confirmed by DSP had increased expression of T cell markers. Tumor proliferation (p < 0.001), hypoxia (p = 0.04), glycolysis (p < 0.001), B7-H3 (p = 0.007) and TGF-β (p = 0.001) signatures were significantly associated with shorter OS. Additional DSP profiling of these mesothelioma samples showed both T cell excluded and desert TMEs. Conclusions: We show that the mesothelioma TME has distinct immune biology associated with OS. Tumors from patients with poor survival had expression profiles previously described to be associated with immune excluded and desert phenotypes. We show that gene expression and DSP identifies unique targets for immunotherapy and we hypothesize that these findings may guide the development of combination trials that will be effective against mesothelioma.