Identification of Subtype-Specific Vulnerabilities in Resistant Glioblastoma: A Computational Pipeline for Biomarkers and Drug Discovery

Therapeutic resistance
in glioblastoma (GBM) remains
a critical
clinical challenge. To better understand its molecular basis, we performed
a multistage computational analysis on a resistant GBM data set derived
from the Ivy Glioblastoma Atlas Project, filtered for unmethylated
MGMT and EGFR amplification. A subtype-specificity analysis identified
distinct expression patterns for established biomarkers, including
PDGFRA, FAP, and CD163, providing refined context for their roles
in specific GBM subtypes. Kaplan–Meier analysis confirmed that
high PDGFRA expression correlates with poor survival in this resistant
population. To explore its therapeutic tractability, we developed
a Quantitative Structure–Activity Relationship (QSAR) model
targeting PDGFRA. This led to the in silico identification
of a novel lead compound with a distinct thiazolopyridine-based scaffold
and high predicted potency. Our findings demonstrate how integrated
bioinformatic pipelines can dissect the complex landscape of GBM resistance,
contextualize the roles of key oncogenes, and guide the rational design
of potential new inhibitors.