Understanding the role of tumor microenvironment in low grade glioma progression to malignancy - ABSTRACT
Our PCA research center brings together the UCLA Neurosurgery team with the Caltech spatial single cell team
to construct a comparative spatial atlas of low-grade gliomas. The collaboration between the three components
of the PCA research center has already generated preliminary data in several glioma samples. We will expand
this effort to generate a comparative atlas of gliomas with distinct progression outcomes using integrated spatial
transcriptomics, proteomics, and chromosome profiling. By comparing the low-grade gliomas that eventually
transform with ones that stay indolent or do not recur, and with IDH-mutant high-grade gliomas, we aim to
understand the molecular and cellular mechanisms at the low-grade stage that are predictive of malignant
transformation (MT) and to suggest intervention strategies to prevent MT. The comparative analysis will examine
three types of changes in low-grade gliomas with different outcomes: cell type composition, tumor
microenvironment, and pathway specific gene expression. From UCLA’s Brain Tumor Translation Resource
(BTTR) center, we have already collected 99 fresh-frozen low-grade glioma samples and will collect
approximately an additional 100 samples of low-grade glioma with different outcomes (MT, indolent, and no-
recurrence). 38% of the current cohort of patients are from under-represented minority groups; we will continue
to recruit from a diverse patient pool in order to better understand which patients may be at higher risk for
malignant transformation and therefore need more frequent surveillance or earlier intervention. We will then
generate an integrated multi-modal spatial atlas targeting 2500 mRNAs, 10 proteins and 10 DNA CNVs and
translocations. From the high sensitivity and multiplexed RNA seqFISH assays, we will be able to capture not
only cell type and microenvironment information, but also genes and pathways that could be causal for
progression to malignancy. Lastly, we will use the data to 1) predict tumor progression based on the cell type
compositions and microenvironments; 2) design intervention strategies based on the spatial data, using
counterfactual inference models to affect immune infiltrating and other predictors of progression; and 3) build a
model of tumor progression dynamics based on gene expression and mechanics of the tissue. Our
comprehensive low-grade glioma tissue collection, the integrated spatial dataset with transcriptomics,
proteomics and chromosomal abnormalities, and the models built using advanced machine-learning tools will
extend the existing capabilities of the HTAN consortium and be interoperable. The atlas and the computational
tools will be used by us and the wider scientific community to further understand the mechanisms leading to
malignant transformation.
