DMG-Alt continues to be a devastating disease with limited therapeutic options, however, in the last decade neurosurgical and neuro-oncological techniques have advanced considerably allowing for biopsy and sample collection from these tumors. This in turn has led to several novel clinical trials currently underway based on the genomic targets discovered from comprehensive transcriptomic studies. Our study aims to push the field forward by taking the next step in genomic screening by examining spatial aspects of the tumor architecture as well as proteomic profiles. Spatially, we found that tumor ROIs that contain portions of TME display distinct genetic signatures compared to regions with dense tumor cell populations. Using proteomics, we demonstrate that transcription to translation fidelity is not always maintained in both adult and pediatric DMG-Alt. Finally, we interrogated common genetic targets proposed both for these tumors as well as broadly used in other gliomas and reveal the shortfalls of predicting therapeutic targeting efficacy based on RNA data alone. Continued research in this field will be crucial in advancing both the molecular and clinical understanding of these rare and lethal tumors.
A recent publication by Liu and colleagues demonstrated an increased tendency for pediatric DMG-Alt to be dominated by an OPC like cell type while adult DMG-Alt enriched for a mesenchymal cell phenotype14. In our cohort we saw similar patterns across adult and pediatric tumors, however when examining the spatial landscape, we observed a change in the signature in ROIs composed of both tumor and TME cells. Also of note, in both our cohorts of data there is little to no neuronal signature expressed within any sample set. This is interesting because these tumors arise in very neuronally dense and eloquent tissue but seem to not be as intermixed with their surroundings like glioblastomas (GBM), the most common primary CNS malignancy in adults, which are known to integrate into neuronal synapses and proliferate more in response to neuronal firing43–46. This may indicate that DMG-Alts interacts more with other brain resident cells such as microglia, astrocytes, or oligodendrocytes as opposed to neurons. It’s also worth noting that there was little to no outside immune cell infiltration into the tumor bulk itself which is consistent with reports of DMG-Alt being an immune desert47,48.
Our study is the first to include spatial proteomic profiling and we demonstrate that while translational fidelity was measured at similar levels across adults and pediatric samples, as well as across the ROIs, there were subtle differences in the types of genes that were either faithfully or not faithfully translated from mRNA. These results have significant implications for cancer therapeutics in general which to date has relied mostly on screening mRNA due to difficulties in obtaining proteomic data. Our examination of common genomic targets discovered using mRNA alone shows that although this method can result in true proteomic hits there is also room for error including both false positive and false negative hits being prioritized (such as EGFR false negative vs. PD-1 false positive). Although there are some studies suggesting mRNA, specifically differentially expressed mRNA, may be a faithful readout of true protein synthesis49, this has yet to be widely examined in the context of brain malignancies, especially in the pediatric population.
One of the limitations of our study is that the number of protein probes available is far less than the number of RNA probes, and so a genome-wide comparison of RNA to protein was not possible. We also did not have a large enough number of samples to include statistically significant sex comparisons. In addition, due to the technical limitations of the Nanostring platform regarding the number of antibodies that can be used for defining ROIs, we had a subset of cells that were unstained. Since all our samples were targeted needle biopsies aimed at obtaining the most pathological tumor tissue, we hypothesized that these were most likely tumor cells that lacked the H3K27m mutation. In validating this using ORA analysis, we discovered a unique genetic program within these cells compared to H3K27m tumor cells involving some immune signatures as well as cancer and developmental gene enrichment. Although we only visualize a limited number of immune cells in our samples, there were samples with strong enrichment for immune related signaling, especially when TME ROIs were considered. This may be a result of tumor cells hijacking immune signaling methods (seen in other gliomas50–52) or could result from immune cells sitting mostly on the peripheral edges of the tumor (which would not be captured during biopsy) inducing inflammation. An in depth understanding of the immune response and its resulting inflammation in this context may prove critical as we try to advance therapies, particularly CAR-T cell based therapies which are currently undergoing clinical trials53–55.
Although DMG-Alt remains a 100% lethal tumor with incredibly limited survival, novel research is starting to illuminate its genetic background and expose possible therapeutic vulnerabilities. Further studies across multiple institutions to increase sample size will be the key in validating and advancing effective therapies and ultimately making a difference in patient survival.