Background
Diffuse midline glioma, H3 K27-altered (DMG-Alt) are highly aggressive malignancies of the central nervous system that primarily affect the pediatric population. It is classified as a World Health Organization grade 4 neoplasm with no curative treatments. Large scale spatial transcriptomic studies have implicated that tumor microenvironmental landscape plays an important role in determining the phenotypic differences in tumor presentation and clinical course, however, data connecting overall transcriptomic changes to the protein level is lacking.
Methods
The NanoString GeoMx Digital Spatial Profiler platform was used to determine the spatial transcriptomic and proteomic landscape in a cohort of both pediatric and adult H3 K27-altered DMG biopsy samples. Three fluorescently labeled antibodies targeting immune cells (CD45), epithelial cells (PanCK), tumor cells (H3 K27M) and a nucleic acid stain (SYTO-13) were used to establish regions of interest (ROI) for genomic and proteomic analysis including tumor microenvironment (TME), H3K27M tumor cells (Mutated) and tumor cells without H3K27M mutation (Non-Mutated) (figure 1).
Results
Genomic profiling using the SCENIC algorithm demonstrated different regulons driven by ARID3C and MSX1 in the TME, TBP and ELK1 in Mutated, and NFATC2 and Stat6 in Non-mutated ROIs. These regulons were enriched for cytokine and interleukin signaling in the TME ROIs (FDR <0.05), cancer induced senescence and TP53 gene regulation in the Mutated ROIs (FDR <0.10), and TNF and WNT pathway signaling in Non-Mutated ROIs (FDR <0.05). Leveraging the unique Nanostring GeoMx technology we detected both RNA and protein probes across our entire range of samples and discovered although similar rates of RNA to protein correlation on average in adults and pediatrics (0.33 +/- 0.25 SD) were present, the sets of genes that were correlated were highly varied. We also examined common clinical trial targets in DMG-Alt for RNA to protein correlation. PD1, PDL-1 and CTLA-4 expression showed poor concordance between RNA and protein across all the above markers with CTLA-4 especially enriched for negative correlation when examined.
Conclusions
Together these results indicate that DMG-Alt tumors are genomically diverse not only across patient age but also within spatial regions of the tumor. Furthermore, our data indicate RNA to protein translational fidelity in DMG-Alt is not uniform and should be taken into account when selecting promising targets for drug or therapeutic development. These data may begin to shed light on why these tumors driven by identical mutations have drastically different outcomes in the pediatric population compared to the adult population.
Ethics Approval
All patients analyzed within the study provided informed consent and all analyses were carried out according to UWSMPH/UW-Madison IRB direction and supervision. All samples were obtained from biopsies or resections done pre-treatment at the time of initial diagnosis. All analysiss conducted on human samples was performed in accordance with relevant guidelines and regulations.
Abstract 181 Figure 1
Patient cohort and experimental workflow: A) Graphic depicting patient information. B) Graphic depicting the GEOMX slide layout. C) experimental workflow and ROI definition. D) MRNA to Protein correlation mismatch in cancer vs normal cells