Transcriptomic determinants of peptide-drug conjugate sensitivity in patient-derived glioblastoma models

Sammanfattning

Glioblastoma multiforme (GBM) is the most aggressive primary brain tumour in adults, with a median overall survival of approximately 15 months despite standard-of-care therapy. Peptide-drug conjugates (PDCs) are a new generation of therapeutics which rely on high levels of tumour aminopeptidase activity to selectively release a cytotoxic payload (alkylating agent) and predictive biomarkers of response to these PDCs are not yet defined. The objectives of this thesis were to identify transcriptomic determinants of PDC sensitivity in patient-derived GBM models and to validate the resulting signature in an independent solid-tumour validation cohort. Two biobank data sets were used for a fully computational reanalysis: (1) a biobank of Human Glioblastoma Cell Culture (HGCC) with matched expression and IC₅₀ data for two PDCs across 48 patient-derived cell cultures, and (2) a 16-model breast cancer organoid panel from Crown Bioscience with matched Compound B IC₅₀ data. Gene–IC₅₀ Spearman correlations and over-representation analysis in HGCC and gene set enrichment analysis in organoids were used together. Both compounds converged on a near-identical single-gene response signature: the strongest single-gene correlate of sensitivity for both compounds was ZNF121, together with genes involved in mitotic-segregation machinery (KIF20B, MPHOSPH9) and structure-resolution factors of the homologous-recombination axis (BLM, RMI1). Cell-cycle and chromosome-segregation processes were the most enriched pathways. Cross-platform validation of the signature in breast cancer organoids was performed and confirmed the signature: BRCA1, BRCA2, BLM and RAD51 demonstrated directional concordance and organoid GSEA reflected the cell-cycle biology observed in HGCC. An extracellular-matrix and stromal-signalling resistance signature was also seen for Compound A. These results highlight a replication-and-repair-coupled transcriptomic vulnerability that is shared across two structurally distinct PDCs and conserved across tumour type and dimensionality, providing candidate biomarkers for future functional validation and clinical translation.