Paired bulk transcriptomic analysis of recurrent glioblastoma with primary single-cell reference mapping.

Glioblastoma (GBM) is the most aggressive primary brain tumor in adults, and virtually all patients relapse on standard temozolomide (TMZ) plus radiotherapy. The molecular mechanisms that allow residual tumor cells to survive treatment, reshape the microenvironment, and seed a drug-resistant recurrence remain incompletely understood. We analyzed paired bulk RNA-seq from 13 primary-recurrent patient pairs (GSE139533), used two published primary GBM single-cell/single-nucleus samples (GSE138794/GSM4119534 and GSE174554/GSM5319503; 5244 quality-controlled cells/nuclei) solely for cellular-context mapping, and evaluated an exploratory bulk-derived gene set in TCGA-GBM. Paired testing of 18,445 genes identified one significant gene at FDR 〈 0.05 and |log2FC| 〉 0.3: IL4I1 (log2FC = 1.49, FDR = 0.023). A 40-gene exploratory set was defined as the 40 positive paired-effect genes with the strongest paired-test evidence in the same 13 pairs and used descriptively; its mean paired score difference was 1.02 z-score units, but no independent significance test was assigned to this within-cohort score. In the primary single-cell references, the set scored highest in Myeloid-like cells, providing cellular context without constituting a primary-recurrent single-cell comparison. TCGA-GBM showed a weak, non-significant baseline survival association (Cox HR 1.30, 95% CI 0.91-1.87, P = 0.15). A TMZ-resistant U251 derivative had a higher TMZ IC50 than parental cells (612 vs 89 µM), and Western blotting showed increased MGMT, CD44, Vimentin, HIF-1α, OPN and PD-L1 across three biological experiments. Together, these data provide a paired bulk description of recurrence-associated expression and use primary single-cell references only to localize the exploratory signal.