INTEGRATING TRANSCRIPTOMICS WITH SPATIAL RADIOGRAPHIC ATLASES FOR GLIOMA ANALYSIS

Abstract AIMS The localization of gliomas in the brain is a critical prognostic factor, reflecting the genetic makeup of their originating cells. Variations in transcriptomic profiles across different regions and subtypes may further shed light on the mechanisms driving tumor development. Our research delves into the complex interplay between glioma locations and their transcriptomic signatures, seeking to dissect the spatial and genetic intricacies contributing to glioma diversity. METHOD In this study, we conducted a comprehensive analysis of preoperative anatomical MRI and transcriptomic data from 242 glioma patients to investigate the relationship between spatial distribution and tumor transcriptomic profiles. The MRI data were standardized by registration to the MNI-152 brain space. Tumors were segmented, and the brain was partitioned into 84 discrete subregions for detailed examination. Transcriptomic data underwent preprocessing, with principal component analysis employed to reduce dimensionality. Statistical analyses were used to assess the relationship between glioma description across different subtypes and PCA results. Additionally, we identify differentially expressed genes and common genes across regions for various glioma subtypes. Comparative analyses with the Allen Brain Atlas (divided into 84 subregions with expression data) were also performed to discern transcriptional differences between healthy individuals and glioma patients, further enhancing our understanding of glioma biology at the molecular level. RESULTS We generated 21 frequency maps detailing regional glioma occurrences, categorized by biomarkers, subtypes, and grades. Notably, some of these maps showed significant correlations with expression principal component analyses (PCAs). Our findings include a set of genes consistently expressed across subregions, alongside region specific differentially expressed genes. Additionally, we identified distinct gene expression patterns when comparing glioma patients to healthy controls in the Allen Brain Atlas, highlighting potential molecular underpinnings of glioma heterogeneity. CONCLUSION Our analysis revealed notable correlations between glioma subtype locations and transcriptomic profiles, underscoring the role of transcriptomic signatures in explaining regional variations in glioma incidence.