Bortezomib (proteasome inhibitor; NF-κB/MGMT suppression) + temozolomide shows 24% RANO responses in recurrent MGMT-unmethylated GBM—biomarker signal tied to HSPBP1 insertion + intact apoptosis/NF-κB

LB014 reports a biomarker-driven readout from the ongoing Phase Ib/II NCT03643549 trial testing sequential bortezomib (a 26S proteasome inhibitor that can blunt NF-κB survival signaling and reduce MGMT) followed by temozolomide in recurrent MGMT-unmethylated glioblastoma (n=58; training n=43, prospective validation n=15; 116 mpMRIs with longitudinal deep-learning segmentation). Objective RANO responses occurred in 14/58 (24%); a prospective multimodal ML model achieved AUC 0.91 in validation (permutation p=0.026), with an HSPBP1 9-bp insertion the dominant predictor (present in 100% of responders vs 59% of non-responders; p=0.0027, OR>100), but responses required preserved downstream pathway function (responders had lower loss-of-function burden in apoptosis, NF-κB, cell-cycle, and RTK pathways; all p<0.03). Compared with commonly used recurrent-GBM benchmarks, this 24% ORR is in the ballpark of regorafenib’s disease-control-driven activity (REGOMA DCR 44% with mOS 7.4 vs 5.6 months vs lomustine) and below the higher radiographic response rates often reported with bevacizumab (≈20% ORR in early studies; higher in some retrospective series), while the key differentiator here is the proposed predictive genomic/radiogenomic enrichment strategy rather than an across-the-board efficacy claim.