CheckMate 143

Background

CheckMate 143 was a phase III randomized open-label trial evaluating nivolumab (anti-PD-1) versus bevacizumab in patients with GBM at first recurrence after standard temozolomide-based chemoradiation. At the time of design, bevacizumab was considered the standard of care for recurrent GBM based on response rate data and PFS improvement (BRAIN, EORTC 26101), though no OS benefit had been formally demonstrated. CheckMate 143 tested whether PD-1 blockade could improve upon bevacizumab outcomes in the recurrent setting, given the hypothesis that immunotherapy would produce more durable responses.

Interventions and follow up

Arm A: Bevacizumab 10 mg/kg IV q2w until progression
Arm B: Nivolumab 3 mg/kg IV q2w until progression
Primary endpoint: Overall survival (OS); secondary: PFS, ORR, safety, QoL
mFollow up: 9.5 months (OS primary); 28 months (updated analysis)

Results

mOS: 9.8 months (nivo) vs 10.0 months (bev), HR 1.04, 95% CI 0.83–1.30, P=.83 (no difference)
mPFS: 1.5 vs 3.5 months, HR 1.97 (favoring bevacizumab)
ORR: 7.8% (nivo) vs 23.1% (bev)
MGMT-methylated subgroup mOS: 10.4 (nivo) vs 12.9 months (bev)
Biomarker analysis: no biomarker (PD-L1, TMB, MGMT) predicted nivolumab benefit

Adverse events

Grade ≥3 AEs: 33.2% (nivo) vs 35.0% (bev)
Immune-mediated AEs (nivo, all grades): fatigue 25%, headache 23%, rash 10%, hypothyroidism 9%
Grade ≥3 immune-mediated (nivo): 14%
Bevacizumab: hypertension 7.6%, thromboembolic events 8%, GI perforation 2%
Deaths: no treatment-related deaths in either arm
Corticosteroid use: 28% of nivolumab patients for immune-related events

Conclusions

Nivolumab did not improve OS compared to bevacizumab in recurrent GBM (mOS 9.8 vs 10.0 months, HR 1.04), with lower response rates and shorter PFS, establishing that single-agent PD-1 checkpoint blockade is not effective in unselected recurrent GBM and suggesting that the immunosuppressive tumor microenvironment of GBM represents a major barrier to immune checkpoint inhibitor efficacy.

Key Limitations

Open-label design; bevacizumab comparator itself has no OS benefit vs lomustine in recurrent GBM (EORTC 26101), making the control arm selection debatable; GBM is characterized by profound immunosuppression (T-cell exclusion, high M2 macrophage infiltration, TGF-β overexpression, IDO expression) limiting ICI efficacy; low TMB in most GBM; steroid dependence at recurrence (often high-dose dexamethasone) profoundly suppresses immune function and likely abrogates any ICI benefit; no biomarker (PD-L1, TMB, IDH, MGMT) predicted benefit; hypermutated GBM after TMZ treatment (~10%) was not prospectively enriched; combination immunotherapy (nivo+ipi) failed similarly in CheckMate 498/548.

Clinical Context

CheckMate 143 definitively demonstrated that single-agent PD-1 blockade is ineffective in unselected recurrent GBM, a finding subsequently confirmed by CheckMate 498 (nivo vs TMZ in MGMT-unmethylated newly diagnosed GBM, no benefit) and CheckMate 548 (nivo+TMZ in MGMT-methylated newly diagnosed GBM, no benefit). GBM remains one of the most immunotherapy-resistant tumors. Bevacizumab remains a commonly used recurrent GBM agent in practice despite no OS benefit. Standard recurrent GBM options include lomustine, bevacizumab, re-irradiation, and TTFields (EF-11 data). Ongoing trials explore personalized vaccines (DCVax-L), CAR-T cells, oncolytic viruses, and EGFRvIII-targeted therapies.

References