Professor Yale University New Haven, Connecticut, United States
Introduction:: Glioblastoma is the most common primary malignant brain tumor, associated with poor prognosis and debilitating effects on the patient’s quality of life. Though glioblastoma develops from astrocytes, glioblastoma tumor cells come in a variety of subgroups that can affect patient treatment and survival. Though the genotypes of glioblastoma subgroups have been widely studied, the phenotypes have been less extensively studied. In this work, we use co-detection by indexing (CODEX) to obtain spatial proteomic data on glioblastoma tissue, and demonstrate its co-registration with multimodal stimulated Raman scattering (SRS) images. Furthermore, we present 46-plex spatial protein data of glioblastoma subtypes present in a single tumor.
Materials and Methods:: Two samples of human formalin-fixed paraffin-embedded (FFPE) glioblastoma tissue sections with two adjacent slides each were imaged. For each sample, one slide was imaged via stimulated Raman scattering (SRS), second harmonic generation (SHG), and two-photon microscopy, then both slides were imaged using the PhenoCycler-Fusion system. The system uses DNA-conjugated antibodies, each with an assigned barcode, to target a large array of biomarkers. Corresponding fluorescent reporters are iteratively added, imaged, and removed to image three markers each cycle. For this study, 46 antibodies were used to profile the immune state of the glioblastoma tumor environment. After data collection, cell segmentation was performed using StarDist. Cell clustering and gene expression analysis was carried out using Seurat.
Results, Conclusions, and Discussions:: A 46-plex panel of immune and tissue markers were successfully imaged on glioblastoma tissue. Using distinct features such as tissue structure, collagen fibers, and vasculature, the multiplex images were superimposed on optical data, and demonstrated the capability of co-analysis between the two modalities to provide a more comprehensive understanding of the disease. We also identified different subtypes of glioblastoma within the same tissue, distributed in a ring-like manner. Downstream analysis revealed differences across certain regions, with varying levels of cell proliferation and leukocyte infiltration that reflected this ring-like structure. We hope that further analysis of the multiplex data, combined with metabolic data from the multimodal optical images, will reveal more about the spatial phenotypic profile of these glioblastoma subtypes.
