REGORAFENIB IN GLIOBLASTOMA: MOLECULAR CHARACTERIZATION OF TUMOR RESPONSE

Glioblastoma, IDH-wildtype (GBM) is a highly malignant brain tumor endowed with a dismal prognosis (median overall survival 15 months). After bulk tumor removal and standard chemoradiation with temozolomide, tumor invariably recurs leading to patient death. Tumor recurrence has been linked to the presence of glioma stem-like cells (GSCs), which are largely resistant to treatments. Recently, the phase II REGOMA trial demonstrated the effectiveness of the multikinase inhibitor regorafenib (REGO) on GBM relapsed patients, paving the way to REGO incorporation in most guidelines. Despite the introduction in the clinical practice, molecular mechanisms driving REGO responsiveness are still poorly understood. Here we present a CRISPR/Cas9 chemogenomic screening-based method to identify REGO-synergistic drugs in human GSCs. METHODS For the CRISPR/Cas9 screening, we took advantage from the minimal genome-wide human gRNAs library (MiniLibCas9), which allows the use of a reduced number of gRNAs targeting the whole genome. GSCs were firstly transduced by lentiviral infection with Cas9 enzyme. Cas9-expressing GSCs were infected with the MiniLibCas9 lentivirus at a low multiplicity of infection, to ensure that, statistically, each cell is transduced with a single gRNA resulting in one silenced gene per cell. Stably infected MiniLibCas9 cells have been exposed to vehicle or REGO, with an LD20 concentration, for the time needed for each cell line to carry out 5 or 14 population doublings. Finally, the DNA extracted from GSCs has been screened by Next Generation Sequencing (NGS) to determine the relative abundance of gRNAs in the population treated with regorafenib vs control. RESULTS For each screen, we pinpointed genes surpassing the 5% FDR threshold and designated them as significantly essential. Subsequently, we filtered out core-fitness genes using a comprehensive collection of prior-known essential genes and considered those exclusively deemed essential in the CRISPR screens upon REGO treatment (in comparison to respective untreated controls), and shared among cell lines, resulting in a highly reliable and consensual list of essential genes (N=103). To further prioritize this list, we used two different approaches. The first is based on the druggability of candidate genes, selecting for compounds already approved in clinical practice, targeting at least one of the 103 genes. In the second approach, we functionally mapped genetic vulnerabilities and signatures associated with REGO treatment, by systematic analysis of a second-generation map of cancer dependencies (cMap), recently updated in Pacini C. et al 2024 (PMID: 38215750). By this approach, we identified drugs which collectively down-regulate the set of genes whose expression is necessary for cell survival in the presence of REGO. Both approaches allow the identification of specific drugs to be used as combinatorial treatment to improve REGO effectiveness in GBM. CONCLUSION We confirm the reliability of the CRISPR/Cas9 chemogenomic screening in the study of GSC response to drug treatment. Our integrated high throughput data are currently being validated in preclinical models.