FLASH Modulation of Immune Microenvironment in Diffuse Midline Glioma (DMG) at Single-Cell Resolution

Abstract

<h3>Purpose/Objective(s)</h3> Diffuse Midline Glioma with H3K27M mutation (DMG) is a universally fatal tumor primarily treated with radiotherapy (RT) and typically unresponsive to immune-based therapies. FLASH is a novel RT technique associated with decreased toxicity and effective tumor control in glioma models as well as immune-sparing properties in other cancer models. However, the effect of FLASH on the DMG tumor immune microenvironment (TIME) has not yet been explored. <h3>Materials/Methods</h3> Using a syngeneic murine brainstem DMG, we performed single-cell RNA sequencing on CD45+ cells isolated from tumors 4 days after posterior fossa irradiation (15Gy x 1, 9MeV), utilizing either ultra-high dose rate (90Gy/sec, FLASH) or conventional (2Gy/min, CONV) RT, and compared to unirradiated tumor (SHAM). For differential gene expression analysis, we compared CONV to SHAM (CvS) and FLASH to SHAM (FvS). We also performed single cell gene set enrichment scores (ssGSEA) for evaluation of immune signatures from gene ontology (GO); ssGSEA scores were deemed significant by a p<0.05 using Mann-Whitney U test. We validated our most robust immune markers with dual immunofluorescence. <h3>Results</h3> Unsupervised clustering of 23,165 CD45+ cells revealed 16 immune clusters, including subtypes of resident microglia (MG, 78.2% of total cells), lymphocytes (6.4%), macrophages (MPs, 5.2%), monocytes (1.8%), and dendritic cells (DCs, 1.7%). MG consisted of four distinct subtypes representing a spectrum of homeostatic to activated cellular states. In the most activated MG subtype, both CONV and FLASH showed an upregulation of interferon (IFN) genes such as Isg15 (Fold Change (FC) 1.03 CvS, 0.94 FvS) and Ifit3 (FC 0.77 CvS, 0.66 FvS) when compared to SHAM. Furthermore, ssGSEA of MG found IFN scores increased in both FvS (p < 0.0001) and CvS (p < 0.0001) while TGFbR signaling score was increased in FvS (p < 0.0001) and decreased in CvS (p < 0.0001). In non-resident myeloid clusters, the most differential response was found in MPs and DCs with robust upregulation of IFN genes for CvS (Isg15 FC 2.30 MPs, 2.67 DCs; Ifit3 FC 2.02 MPs, 2.06 DCs), while this response was not seen with FvS. This was confirmed with ssGSEA with increased IFN scores in both MPs (CvS p<0.0001, p=0.04) and DCs (CvS p<0.0001, FvS p=0.69). FLASH showed an increase in anti-inflammatory MP markers such as Mrc1, Cd163, and Maf (FC 1.40, 1.11, 1.19) and an upregulated myeloid-derived suppressor cell signature in monocytes (FvS, p < 0.0001), not seen in CvS (p = 0.09). <h3>Conclusion</h3> Our work is the first to map CONV and FLASH immune alterations with single-cell resolution in the DMG TIME. FLASH induces IFN differentially across microglia and non-resident myeloid subsets. Further exploration is warranted to understand how FLASH impacts tumor growth through differential cytokine signaling, inflammatory responses and immune checkpoint expression within the DMG TIME.