Targeted drugs plus radiotherapy fail to extend survival in children with DIPG brain tumors
The BIOMEDE randomized trial of 233 patients found no survival gain from erlotinib, dasatinib or everolimus, but TP53 mutation strongly predicted outcome.
Summary
Diffuse intrinsic pontine glioma (DIPG) is an incurable brainstem tumor that mostly strikes children. The BIOMEDE trial randomized 233 patients, with a median age of about 8 years, to radiotherapy plus one of three targeted drugs: erlotinib, dasatinib or everolimus. Assignment was guided by tumor biomarkers. None of the drugs improved overall survival compared with a historical control group: median survival was 9.7, 9.9 and 11.9 months, versus 10.8 months in controls. The trial stopped early for futility. Everolimus caused fewer side effects and fewer treatment stops because of toxicity. Tumors with TP53 mutations did much worse (hazard ratio 2.8). Four long-term survivors treated with an mTOR inhibitor were alive and off treatment more than 6 years after diagnosis. The data give future trials better biomarkers to build on.
Detailed Summary
Diffuse intrinsic pontine glioma (DIPG), now defined as diffuse midline glioma with loss of H3K27 trimethylation, is the pediatric tumor with the worst prognosis. Surgery is impossible, radiotherapy only delays progression, and no chemotherapy has helped. Targets such as EGFR, mTOR and PDGFRA, found through biopsies and autopsies, prompted this trial of biology-guided treatment.
BIOMEDE was an international randomized phase 2 trial that enrolled 326 patients from 2014 to 2020. All had biopsy-proven DIPG, and the trial was led largely from France. Randomization depended on tumor biomarkers: erlotinib could only be assigned if EGFR was overexpressed, and everolimus only if PTEN was lost. In total, 233 patients were randomized (36 erlotinib, 95 everolimus, 102 dasatinib) to receive one drug with 54 Gy radiotherapy and then as adjuvant therapy until progression. The primary endpoint was overall survival. A control cohort of 66 biopsy-proven patients treated earlier, mostly with temozolomide, served as a comparator.
No drug improved survival. With median follow-up of 5.3 years, median OS from biopsy was 11.1 months in the trial and 10.8 months in controls. By arm it was 9.7 months (erlotinib), 9.9 (dasatinib) and 11.9 (everolimus). Head-to-head hazard ratios were 0.87 for erlotinib versus dasatinib, 0.94 for erlotinib versus everolimus and 0.89 for everolimus versus dasatinib (95% CI 0.66-1.19, P=0.42). Median progression-free survival was 6.4 versus 7.1 months for everolimus versus dasatinib. The independent data monitoring committee stopped randomization for futility in September 2019.
Safety differed between arms. Treatment was stopped for toxicity in 20% on erlotinib, 14% on dasatinib and 3% on everolimus (P=0.004). Severe skin events were more frequent with erlotinib, and severe renal and gastrointestinal events with dasatinib. Metabolic events were more common with everolimus. No treatment-related deaths occurred, but 78% of patients had grade 3-4 adverse events. Biopsy was feasible, with no biopsy-related deaths, though 27 of 157 centrally reviewed procedures were followed by neurological worsening within 48 hours and 14 severe complications were reported.
Exploratory genomics on about 172 tumors found TP53 mutation in 57%, which was the strongest prognostic factor (multivariable HR 2.84, 95% CI 1.92-4.20, P<0.0001). TP53 mutation clustered with H3.3K27M and with many structural chromosomal changes. Histone subtype lost prognostic value after adjustment. Mutations in or activation of the mTOR pathway were associated with better response to everolimus, and four long-term survivors on mTOR inhibitors were alive and off treatment beyond 6 years.
The main result is a clear null: none of these three agents with radiotherapy beat historical outcomes. The mTOR signal is hypothesis-generating only. It comes from exploratory subgroup analyses and four survivors, without a randomized comparison against standard care. Everolimus's better tolerability makes it a reasonable backbone for combination trials, now with a defined molecular risk stratification.
Key Findings
- Primary endpoint not met: median OS 9.7 months (erlotinib), 9.9 (dasatinib), 11.9 (everolimus) versus 10.8 months in the historical control cohort
- Everolimus versus dasatinib OS hazard ratio 0.89 (95% CI 0.66-1.19, P=0.42) in 193 patients; PFS 6.4 vs 7.1 months, no difference
- Treatment stopped for toxicity in 20% (erlotinib), 14% (dasatinib) and 3% (everolimus), P=0.004; 78% overall had grade 3-4 adverse events
- TP53 mutation (57% of tumors) was the strongest predictor of poor survival: multivariable HR 2.84 (95% CI 1.92-4.20, P<0.0001)
- Mutations in or activation of the mTOR pathway were associated with better everolimus response (exploratory); four mTOR-inhibitor-treated long-term survivors were alive and off treatment beyond 6 years
- Biopsy had no deaths, but 27 of 157 centrally reviewed procedures had neurological worsening within 48 hours and 14 had grade 3-4 complications
- Randomization closed for futility after 233 patients; 49% had pseudoprogression and 75% had clinical improvement, with no difference between arms
Methodology
BIOMEDE was an international randomized phase 2 trial (NCT02233049) in biopsy-proven H3K27-altered DIPG, with biomarker-conditional randomization (R1, R2, R3) among erlotinib, everolimus and dasatinib given with 54 Gy radiotherapy. The primary endpoint was OS by intention to treat using Kaplan-Meier and stratified log-rank tests; 233 patients were randomized. A historical cohort of 66 biopsy-proven patients, mostly treated with temozolomide, was a comparator. Whole-exome and RNA sequencing were exploratory, analyzed in about 172-176 tumors.
Study Limitations
The control group was historical, from a single center, and mostly treated before the trial, so comparisons are non-randomized. Biomarker-restricted randomization produced unequal arm sizes (only 36 on erlotinib), underpowering some comparisons, and post-progression treatment was not standardized, with 25% reirradiated. Biomarker and genomic findings, including the mTOR signal and the four long-term survivors, were exploratory and not adjusted for multiplicity.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe:
