Journal of Medical Sciences and Health Care Research
BBB-Gated h-Foci Map State-Resolved Anchors for Cell Therapy and Targeted Protein Degradation in Glioblastoma
Abstract
Fred Y. Ye
Computational drug repurposing in glioblastoma is vulnerable to cell-composition confounding, lineage-mismatched perturbation models and unmeasured blood-brain barrier exposure. We integrated three case-control cohorts (247 tissues) to define 4,762 directionally concordant genes and applied continuous h selection to extract a 293-node differential network with three h-foci and 54 second-order hubs. Donor-aware analysis of 21,943 cells localized these programs across Neftel states without treating tumour-only data as a case-control comparison. Among 53 perturbagens with active Drugs@FDA products scored in U251MG and LN229—the only represented GBM lines—none showed reproducible reversal in both lines, and every relaxed signal failed a conservative CNS descriptor screen. Afatinib and temozolomide served as complementary controls for sufficiency and necessity. This null result redirected the network into a spatial modality atlas: CD276 and IL13RA2 prioritized cell-therapy tests, whereas ROCK1, DBF4, ODC1 and CCNA2 prioritized targeted protein degradation. These are falsifiable experimental anchors, not efficacy claims.

