Skip to main content
2024 AANS Annual Scientific Meeting Main banner
Back

Neurosurgeon Forum

Tumor

Integrated Multi-omic Spatial Reconstruction of the IDH-Mutant Glioma Microenvironment

Friday, May 3, 2024
1:20 PM – 1:24 PM CT
Location:  S401ab
Disclosure(s):
Danielle F. Miyagishima: No financial relationships to disclose
Jennifer Moliterno-Gunel, MD: Stryker: Consultant (Ongoing)
Introduction: Tumor heterogeneity poses a considerable challenge in treating various tumor types, including gliomas, hindering the effectiveness of non-surgical interventions. We employed multi-omic integration of single-cell RNA sequencing (sc-RNAseq), spatial transcriptomics, and computational histopathology, applying a novel spatial denoising algorithm to reveal intratumoral signaling axes and specific cell types associated with IDH-mutant (IDH-m) glioma progression.

Methods: We analyzed 38,228 whole-transcriptome spatially barcoded spots from 11 surgically excised human astrocytic IDH-mutant gliomas, including WHO grade 2 (n=3), grade 3 (n=3), and grade 4 (n=5). This analysis incorporated data from two previously published cohorts comprising 16,941 single cells from 9 IDH-m grade 2-4 astrocytic gliomas and validated using TCGA bulk RNA sequencing. We then used the Spatial Affinity-graph Recovery of Counts (spARC), a novel manifold-learning based algorithm to denoise raw spatial transcriptomic mRNA expression, to simulate receptor-ligand interactions within the tissue architecture.

Results: We identified spatially defined tumor niches and gene markers, including CD44, VIM, LDHA (dense tumor); SNAP25, UCHL1, OLMF1 (leading edge); OLIG2, MBP, PDGFRA (infiltrating/high tumor burden). Histological integration demonstrated that regions with high nuclei density correlated with known glioma progression markers, an observation exclusive to spARC application and not seen with raw mRNA expression alone. Additionally, integration with scRNA-seq and fractional abundance estimation revealed an enrichment of proliferative stem-like cells in these regions. Simulating ligand-receptor interactions led to the identification of SPP1-CD44 signaling enrichment in Grade-4 samples, confirmed by independent sc-RNAseq and TCGA cohort validation.

Conclusion : By using a novel comprehensive approach, combining machine learning spatial transcriptomics denoising tool (spARC), in tandem with sc-RNAseq and computational histopathology, we for the first time, revealed a subset of proliferative stem-like cells characterized by an expression profile previously associated with glioma progression in bulk sequencing, identifying novel therapeutic targets. This approach, employing spatial multi-omic integration, has the potential to identify novel targets in other heterogeneous cancers.