Proteasome inhibitor primes temozolomide response in a molecular subset of recurrent glioblastoma

An interim analysis of a phase IB/II trial testing a sequencing strategy that pairs the proteasome inhibitor bortezomib with the alkylating chemotherapy temozolomide has produced a sharply divided result: the combination achieved disease control in four of ten patients with recurrent, unmethylated-MGMT glioblastoma, while all five patients with grade 4 IDH-mutant astrocytoma in the cohort experienced rapid progressive disease. The findings, published in Biomarker Research, come from the first fifteen consecutively treated patients in the BORTEM-17 trial (ClinicalTrials.gov identifier NCT03643549), recruited between September 2018 and June 2021, and are explicitly framed by the investigators as exploratory and hypothesis-generating.

The therapeutic rationale rests on a well-defined resistance mechanism. Temozolomide kills tumour cells chiefly by adding alkyl groups to the O6 position of guanine in DNA, a lesion that the repair enzyme O6-methylguanine-DNA methyltransferase, or MGMT, can reverse. When the MGMT promoter is methylated, expression of the enzyme is silenced and tumours respond better to the drug. But 50 to 60 percent of glioblastoma patients carry an unmethylated MGMT promoter, which predicts poor response to temozolomide and reduced survival. The research team had previously shown in preclinical work that administering bortezomib 48 hours before temozolomide depletes MGMT mRNA and protein, disrupts pro-survival autophagic flux, reduces NF-κB signalling, and sensitizes unmethylated-MGMT glioblastoma cells to the chemotherapy. On that basis, the trial was designed to test whether the same sensitization could be reproduced in patients.

The treatment schedule reflected that logic. Patients received intravenous bortezomib at 1.3 mg/m² on days 1, 4, and 7 of each 28-day cycle, followed by oral temozolomide beginning on day 3, at doses that were escalated in phase IB from 150 to 200 mg/m² per day for five consecutive days. Eligible patients had histologically confirmed recurrent grade 4 IDH-mutant astrocytoma or IDH-wildtype glioblastoma with an unmethylated MGMT promoter, measurable disease by RANO criteria, radiographic progression at least 12 weeks after radiotherapy, and a Karnofsky performance status of at least 70. The interim cohort of fifteen patients had a median age of 52 years, comprised ten males and five females, and received a median of two treatment cycles, with a total of 42 cycles administered across the group.

The futility safeguard built into the protocol prespecified that the trial would be terminated if none of the first fifteen patients derived clinical benefit, defined as disease control under RANO criteria, meaning either partial response or stable disease. The investigators chose disease control rather than the conventional objective response rate of complete plus partial responses because the enrolled population, recurrent unmethylated-MGMT grade 4 gliomas with dismal prognosis, historically gains limited benefit from temozolomide, and in that context stabilisation of clinical and radiological findings was considered meaningful. By this yardstick the interim signal was positive: overall disease control was 26.7 percent (95 percent confidence interval 7.8 to 55.1 percent), the objective response rate was 6.7 percent, and the prespecified threshold for continuing to full accrual of the planned 63-patient cohort was met.

Crucially, all disease control was confined to the IDH-wildtype glioblastoma subgroup, where four of ten patients benefited: one achieved a partial response, marked by a 42 percent reduction in tumour volume on MRI in a 62-year-old woman, and three maintained stable disease, including one man whose radiological stability persisted for more than eight months. By contrast, every one of the five patients with grade 4 IDH-mutant astrocytoma progressed rapidly. Median time to progression was 4.65 months among responding glioblastoma patients versus 2.00 months for progressing glioblastoma patients and 1.90 months for the IDH-mutant group. Six-month progression-free survival was 25 percent among responders with glioblastoma and zero percent in both progressing subgroups.

Targeted sequencing of glioma-associated genes from tumour DNA revealed a molecular divide that tracked almost perfectly with outcome. Three of the four responding glioblastoma patients harboured EGFR amplification, and all four showed nuclear activation of the NF-κB subunit p65/RelA on immunohistochemistry, with significantly higher nuclear p65 positivity than the rest of the cohort, and none had disruptions of the RB-pathway. EGFR gain was significantly more frequent in patients achieving disease control (p = 0.03; odds ratio 19.98), though the authors caution that the confidence interval, spanning 0.87 to 1691.98, reflects the very small sample. In contrast, every patient who progressed, including all five IDH-mutant astrocytomas, carried alterations affecting the CDKN2A/RB1 cell-cycle regulatory axis, with one IDH-mutant patient additionally harbouring PIK3CA gain. One outlier, an IDH-mutant patient with amplified EGFR and CDKN2A loss who survived unusually long but never achieved a radiological response, suggested that EGFR amplification alone is insufficient when cell-cycle checkpoint disruption or PI3K pathway activation is present.

The authors argue that this pattern may explain the otherwise paradoxical failure in the IDH-mutant group, which normally carries a more favourable prognosis. Loss of CDKN2A removes p16-mediated inhibition of CDK4/6, driving RB phosphorylation and unchecked G1/S cell-cycle transition, while RB1 loss disables the checkpoint more directly, allowing proliferation to continue despite DNA damage. In that setting, even successful MGMT depletion by bortezomib may be bypassed, because tumour cells with collapsed checkpoint control can keep dividing through the temozolomide-induced lesions. The finding is consistent with results from the randomised CATNON trial, in which IDH-mutant patients harbouring CDKN2A/B loss had worse outcomes, and with the incorporation of homozygous CDKN2A/B deletion into the WHO CNS5 classification as sufficient for grade 4 assignment.

Longitudinal plasma proteomics, performed by liquid chromatography-tandem mass spectrometry on samples collected before, during, and after treatment, added a pharmacodynamic layer to the story. Of 1209 proteins detected, 649 differentially expressed proteins were analysed. Patients achieving disease control showed comparatively attenuated treatment-induced changes in proteins linked to invasion and extracellular matrix remodelling, including downregulation of NCAM1 and MMP2, and upregulation of proteins involved in damage clearance and immune modulation, such as AZGP1, CD163, haptoglobin-related protein, and CFHR3. A separate platelet-associated signature emerged: responders preserved plasma levels of cytoskeletal and adhesion proteins including ACTN1, TLN1, THBS1, VCL, and PPBP/CXCL7 during treatment, whereas progressing patients showed marked declines. The authors note that these platelet signals may reflect treatment exposure, host vascular stress, or sample availability differences rather than tumour response itself, and that validation in the full cohort is required.

On safety, the regimen proved feasible. Of 61 adverse events reported, most were grade 1 or 2. The most frequent non-hematological toxicities were fatigue in five patients, constipation and falls in four each, and paresthesia in three. The principal hematological toxicity was thrombocytopenia, affecting three patients, which followed a predictable cyclical pattern with platelet nadirs recovering by day 22 of each cycle, consistent with the phase IB dose-escalation experience. Grade 3 to 4 toxicities were infrequent and considered primarily related to the underlying disease. No dose-limiting toxicities occurred within the four-week assessment window, no patient discontinued treatment due to toxicity, and no grade 5 events or treatment-related deaths were recorded. All fifteen patients ultimately progressed, and all deaths occurred during subsequent follow-up after study withdrawal.

The investigators are careful about the limits of the evidence. The sample is small, causal relationships cannot be established, and plasma proteomics cannot substitute for direct intratumoral assessment, since fresh post-treatment tissue was unavailable because patients were not surgical candidates at progression. The trial also predates the WHO CNS5 classification that formally separated these disease entities. Nevertheless, the interim analysis met its prespecified continuation criteria, and the authors propose that future analyses incorporate EGFR amplification, NF-κB activation, CDKN2A/B status, and RB-pathway integrity as candidate stratification biomarkers. Whole-exome sequencing and integrated proteogenomic analysis of the full phase II cohort, they conclude, will determine whether disease control is reproducibly enriched in tumours with amplified receptor tyrosine kinase and NF-κB signalling and an intact RB checkpoint, potentially enabling biomarker-guided selection of patients for this sequential sensitization strategy.

Subject of Research: An interim analysis of a phase IB/II trial of sequential bortezomib and temozolomide in recurrent grade 4 gliomas with unmethylated MGMT promoter, identifying molecular predictors of response and resistance.

Article Title: Contrasting molecular and pharmacodynamic signatures of response and resistance to sequential bortezomib and temozolomide in recurrent IDH-wildtype glioblastoma and grade 4 IDH-mutant astrocytoma: an exploratory interim analysis of NCT03643549 phase IB/II trial

Article References: Goplen, D., Rahman, M. A., Brekke, J., Waha, A., Oltedal, L., Haasz, J., Kumar, S., Birkeland, E., Selheim, F., Lie, S. A., & Chekenya, M. (2026). Contrasting molecular and pharmacodynamic signatures of response and resistance to sequential bortezomib and temozolomide in recurrent IDH-wildtype glioblastoma and grade 4 IDH-mutant astrocytoma: an exploratory interim analysis of NCT03643549 phase IB/II trial. Biomarker Research, 14(1), Article 148. https://doi.org/10.1186/s40364-026-01023-w

Image Credits: AI Generated

DOI: 10.1186/s40364-026-01023-w

Keywords: glioblastoma, IDH-mutant astrocytoma, bortezomib, temozolomide, MGMT promoter, EGFR amplification, NF-κB, CDKN2A/RB1, proteomics, phase IB/II trial, RANO criteria, drug resistance