MEDB-44. Transcriptomic resolution of subgroup-specific medulloblastoma architecture

Abstract

Despite a growing understanding and stratification of medulloblastoma, it remains an aggressive childhood brain tumor with high morbidity and mortality. Multimodal genomic and epigenomic analysis has permitted the classification of medulloblastoma into four subgroups with varying biology and clinical behavior: WNT, Sonic-Hedgehog (SHH), Group 3, and Group 4. In our previously published work, Single-cell RNA sequencing (scRNAseq) identified distinct tumor cell subpopulations in specific medulloblastoma groups. However, this technology is limited by its lack of architectural information. Spatial transcriptomics is a relatively new technology that permits the analysis of gene expression as it occurs within organized tissue. In our ongoing study, we utilized Visium spatial transcriptomics, integrated with scRNAseq data and immunohistochemistry, to analyze frozen samples of medulloblastomas (SHH, Group 4, and Group 3 with and without MYC amplification). In SHH in particular, we were able to identify scRNAseq populations within the geographically constricted Visium data, including SHH-C2, a population located in histologic nodules, the predominant neuronal-differentiated population SHH-C1, and progenitor populations (SHH-B1 and B2). In addition, we were able to visualize clusters not detectable by scRNAseq – a cluster lining nodules with expression of vascular endothelium marker, reticulin and M2-macrophage genes, and a novel DNA-repair cluster. In addition, Visium data permits the spatial constraint of proliferating cells, which is frequently problematic in scRNAseq, as dividing cells cluster independently. The proliferation is highest in the SHH-B2 minor progenitor population, absent in the SHH-C1 major differentiated population, and is moderate in other population including the SHH-C2 nodules. Group 3 and 4 medulloblastoma are more complex but show preliminary corroboration with scRNAseq data. In summary, Visium allows us to map subpopulations identified by scRNAseq to tumor architecture more definitively and rapidly than IHC. These novel insights advance our understanding of medulloblastoma, a critical step in improving treatment options for children with this disease.