Tissues Harvested Using an Automated Surgical Approach Confirm Molecular Heterogeneity of Glioblastoma and Enhance Specimen's Translational Research Value

Abstract

Glioblastoma (GBM) is the most aggressive primary brain tumor in adults. Designing effective individualized therapies for GBM requires quality fresh tissue specimens, and a comprehensive molecular profile of this highly heterogenous neoplasm. Novel neuro-surgical approaches, such as the automated resection NICO Myriad™ system, are increasingly used by neurosurgeons to better reach the invasive front of tumors. However, no information exists on how harvesting GBM tissue using this approach may impact the translational research value of the sample. Here, we set out to characterize matched specimens from 15 patients, where one tissue sample was obtained using traditional tumor de-bulking (herein referred to as "en bloc" sample), and the other sample was obtained using the Myriad^TM System (herein referred to as "Myriad" sample). We investigated the fidelity of patient derived xenografts (PDXs) for each sample type to the corresponding human tissues and evaluated the added value of sequencing both samples for each patient. Matched en bloc and Myriad samples processed in parallel, were subjected to the following assays: cell viability, self-renewal, in vivo tumorigenicity using an orthotopic model of glioma, genomic sequencing, and pharmacological testing using PI3K-MTOR pathway inhibitors. Our results demonstrate that primary GBM cultures derived from matched specimens grew at similar rates (correlation coefficient R = 0.72), generated equivalent number of neurospheres, and had equivalent tumorigenic potential in vivo (mouse survival correlation coefficient R = 0.93). DNA Sequencing using the Illumina tumor panel amplicons revealed over 70% concordance in non-synonymous mutations between matched human GBM specimens. PDX genomic profiles were also highly concordant with the corresponding patient tissues (>70%). RNA sequencing of paired GBM samples revealed unique genomic variants and differential gene expression between the en bloc and Myriad specimens, with the former molecularly resembling the "tumor core" and the latter resembling the "invasive tumor front" signature. Functionally, we show that primary-derived GBM cells-obtained after fresh specimen's dissociation-are more effectively growth-inhibited by co-targeting non-overlapping mutations enriched in each sample type, suggesting that profiling both specimens more adequately capture the molecular heterogeneity of GBM and may enhance the design accuracy and efficacy of individualized therapies.

Keywords: Myriad; PDX; RNA seq; glioblastoma; tumor heterogeneity.