Uncovering the progressive trajectory of GBM invasion & contralesional neuroplasticity synchrony

218 

Abstract Submission 

Hang Cao¹, Penghu Wei², Yongzhi Shan², Xiaosong He³, Guoguang Zhao²

¹Xuanwu Hospital, Capital Medical University, BEIJING, Beijing shi, ²Xuanwu Hospital, Capital Medical University, BEIJING, Beijing Shi, ³University of Science and Technology of China, Hefei, Anhui

Hang Cao  
Xuanwu Hospital, Capital Medical University
BEIJING, Beijing shi

Penghu Wei  
Xuanwu Hospital, Capital Medical University
BEIJING, Beijing Shi

Yongzhi Shan  
Xuanwu Hospital, Capital Medical University
BEIJING, Beijing Shi

Xiaosong He  
University of Science and Technology of China
Hefei, Anhui

Guoguang Zhao  
Xuanwu Hospital, Capital Medical University
BEIJING, Beijing Shi

Glioblastoma multiforme (GBM) presents significant treatment challenges due to its rapid progression and invasive nature. In vitro studies have demonstrated GBM's capability to propagate excitatory signals through neuron-glioma synapses, facilitating its invasion into adjacent neural networks [2]. However, an in vivo understanding of these tumor-neural interactions is still lacking. Here, we employ virtual brain grafting and the SuStaIn algorithm [3, 4] to tackle the complexities posed by tumor heterogeneity, and hypothesize that by examining contralesional hemisphere neuroplasticity, patients with GBM, despite varying clinical profiles, can be mapped onto a unified model of disease progression.

We enrolled 244 GBM patients, divided into matched left- and right-lesioned subgroups, and 244 matched healthy controls from the UPenn-GBM Project [5] and the Cam-CAN dataset [6]. Pre-operative T1-weighted MRI scans using 3T scanners were conducted. In the GBM groups, lesion segmentation was validated by expert neuro-oncologists and radiologists to exclude cases with bilateral invasion or significant midline shifts. We used FreeSurfer to obtain contralesional DK atlas defined cortical thickness and subcortical volumes. We then quantified contralesional neuroplasticity using Cohen's d map between GBM patients and controls. To assess progressive trajectory, we calculated z-scores for morphometric measures in GBM patients against control norms and used the SuStaIn algorithm's linear z-score model to stage each patient. The primary phases of the trajectory were determined using X-tile's survival cut-off. Last, we conducted phenotype-genotype analyses, starting with phase-correlated gene lists extraction via the GAMBA toolbox, permuting AHBA microarray gene expression data. Gene list annotation utilized Metascape, the cancer single-cell functional state atlas, and oncoEnrichR, focusing on phase-specific pathways and oncological conditions. We also performed pan-tissue cell type annotation using cellKB to examine cell type abundance signature across phases.

In GBM cohort, contralesional neuroplasticity characterized by reduced cortical thickness was observed in both left- and right-lesioned patients. Specifically, trajectories reconstructed with SuStaIn showed progressive cortical thinning starting from the sensorimotor to the limbic system on the cortex, but bidirectional changes in subcortical volumes. No significant correlation was found between the patients' stage with tumor volume, but with age (r's = 0.66/0.71; p's<0.01). Survival analysis suggested a three-phases division, with Phase I showing the best prognosis and subsequent phases showing poorer outcomes (p<0.01). Due to the left-dominance of AHBA data, we performed genetic analysis primarily on the left hemisphere of patients with right GBM. We found that BDNF signaling was the only enriched pathway in Phase I, which was confirmed for influencing neuron-glioma synapse strength and tumor growth [7]. Typical GBM malignancy pathways including WNT signaling, and oxidative phosphorylation pathways were noted across phases, alongside an increase in epithelial-mesenchymal transition activity [8]. Consistently, cell type analysis indicated a shift from astrocytes in Phase I to mesenchymal-type cardiac muscle cells in Phase II. Consistent with Venkataramani et al.'s report that GBM may cause proliferate hyperexcitability through non-synaptic glutamate secretion [9], Phase III showed the highest enrichment for excitatory glutamatergic neurons. Last, compared to other cancers, single cell atlas supported our phase-related genes exhibited GBM-specific functional state changes.

By revealing the alignment between progressive trajectory of contralesional neuroplasticity with known invasive mechanisms of GBM, this study offers a novel perspective on the tumor-neural dynamics and may facilitate the identification of new targets for therapeutic interventions.

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) ¹

Transcriptomics ²

Cortical Anatomy and Brain Mapping

Astrocyte

Glia

Neoplastic Disease

Phenotype-Genotype

Plasticity