What were the key findings of the BIOMEDE trial?

← Back

BIOMEDE

Targeted therapies plus radiotherapy for diffuse intrinsic pontine glioma: the randomized phase 2 BIOMEDE trial

Share Share Copied! Compare

Year of Publication: 2026

Authors: Debily MA, Le Teuff G, Kergrohen T, et al.

Journal: Nature Medicine

Citation: Nat Med 2026;32(6):2201-2215

Link: https://doi.org/10.1038/s41591-026-04354-1

Bottom Line

Adding biomarker-matched erlotinib, everolimus or dasatinib to radiotherapy did NOT improve overall survival in DIPG (median OS ~10–12 months across arms, no different from historical controls), but everolimus showed the best tolerability and signals of benefit in patients with PI3K/AKT/mTOR pathway alterations or 1q gain, supporting it as the standard arm for the next-generation adaptive BIOMEDE trial.

        Major Points
        Largest biology-driven randomized phase 2 trial in DIPG: 233 biopsy-proven patients across 8 countries, randomized to erlotinib, everolimus or dasatinib + radiotherapy based on EGFR overexpression and PTEN-loss biomarkers.
Trial stopped for futility on primary endpoint (overall survival): median OS from biopsy 11.1 months overall vs 10.8 months in historical controls; no pairwise comparison reached significance.
Median OS by arm — erlotinib 9.7 mo (95% CI 7.8–14.6), dasatinib 9.9 mo (8.8–11.2), everolimus 11.9 mo (10.7–14.2).
Everolimus had significantly better safety: treatment discontinuation for toxicity 3% vs 14% (dasatinib) vs 20% (erlotinib), P=0.004; significantly fewer eye, skin and infectious adverse events than erlotinib.
TP53 mutation prospectively validated as the strongest adverse prognostic biomarker (multivariate HR 2.84, 95% CI 1.92–4.20, P<0.0001); H3.1K27M favorable, H3.3K27M worse (driven by TP53 co-occurrence).
Theranostic signal: patients with PI3K/AKT/MTOR pathway mutations, mTOR activation gene-expression signature, or 1q chromosomal gain derived greater benefit from everolimus than dasatinib (median OS 14 vs 9 months).
Four very-long-term survivors (>6 years from diagnosis) were all treated with an mTOR inhibitor.
These results support everolimus as the standard arm for the next BIOMEDE adaptive trial, which will compare everolimus to ONC201.

      

Design

Study Type: Randomized, biomarker-driven, open-label phase 2 trial

Randomization: 1

Blinding: Open-label

Enrollment Period: 2014 – 20 September 2019 (randomization stopped per IDMC recommendation)

Follow-up Duration: Median 5.3 years

Centers: Multicenter international (Gustave Roussy coordinating)

Countries: France, United Kingdom, Denmark, Sweden, Spain, Australia, The Netherlands, New Zealand

Sample Size: 233

Analysis: Intention-to-treat; pairwise comparisons combining biomarker-defined subtrials (R1: erlotinib vs dasatinib; R2: everolimus vs dasatinib; R3: 3-way). Comparison with 66 historical controls (radiotherapy + temozolomide) treated at Gustave Roussy. Independent data monitoring committee.

Registration: NCT02233049

Inclusion Criteria

Clinico-radiologically suspected DIPG with biopsy-confirmed diffuse midline glioma, H3K27-altered, pontine epicenter (or non-brainstem DMG H3K27M-mutant after WHO reclassification)
No prior chemotherapy for the present cancer and no prior cerebral radiotherapy
Age >6 months and <30 years (children <3 years discussed with coordinating investigator)
Eligible for biopsy (or biopsy already performed with material available for biomarker assessment)
Eligible for cerebral radiotherapy
Lansky Play Scale >50% and life expectancy >3 months
Health insurance coverage if required by national regulation
Written informed consent from patient and/or parents/legal representative
Metastatic disease allowed (radiotherapy must start within 3 weeks of biopsy)

Exclusion Criteria

Spontaneous massive intratumoral hemorrhage (controlled postoperative bleeding allowed)
Active intercurrent illness, infection, or uncontrolled comorbidity
Any concomitant anticancer treatment not foreseen by the protocol
Any other cancer within the past 5 years
Previous brainstem irradiation for another neoplasm
Known congenital galactose intolerance, Lapp lactase deficiency or glucose-galactose malabsorption
Pregnancy or breastfeeding
Inability to comply with medical follow-up for geographic, social or psychological reasons
Differential diagnoses on central review (e.g., pilocytic astrocytoma, ganglioglioma, pedHGG-MYCN, MYB-QKI angiocentric glioma, NF1-related glioma)

Arms

Field	Erlotinib + radiotherapy	Everolimus + radiotherapy	Dasatinib + radiotherapy	Control
Intervention	Erlotinib 125 mg/m² once daily during and after radiotherapy 54 Gy normofractionated	Everolimus 5 mg/m² once daily during and after radiotherapy 54 Gy normofractionated	Dasatinib 85 mg/m² twice daily during and after radiotherapy 54 Gy normofractionated	Radiotherapy + temozolomide-based regimen (66 patients with biopsy-proven DIPG treated at Gustave Roussy before BIOMEDE)
Duration	Until progression or unacceptable toxicity	Until progression or unacceptable toxicity	Until progression or unacceptable toxicity	Standard

Outcomes

Outcome	Type	Control	P-value
Overall survival (OS) from biopsy, with pairwise comparisons among treatment arms and vs historical control	Primary	Erlotinib: Median OS 9.7 months (95% CI 7.8–14.6) · Dasatinib: Median OS 9.9 months (95% CI 8.8–11.2) · Everolimus: Median OS 11.9 months (95% CI 10.7–14.2) · Historical Control: Median OS 10.8 months (95% CI 9.5–13.0) · Trial Overall: Median OS 11.1 months (95% CI 9.7–11.7) · Everolimus vs Dasatinib HR: 0.89 (95% CI 0.66–1.19) · Everolimus vs Dasatinib P-value: 0.42 · Erlotinib vs Dasatinib HR: 0.87 (95% CI 0.52–1.46) · Erlotinib vs Dasatinib P-value: 0.59 · Erlotinib vs Everolimus HR: 0.94 (95% CI 0.54–1.65) · Erlotinib vs Everolimus P-value: 0.84 · Primary endpoint NOT met — trial stopped for futility on IDMC recommendation
Progression-free survival (PFS)	Secondary		0.89 (log-rank across all 3 arms)
2-year overall survival distribution	Secondary	144 (62%) survived <1 year; 70 (30%) survived 1–2 years; 19 (8%) survived >2 years; similar across arms
Treatment discontinuation due to toxicity	Secondary		0.004 (Fisher exact)
Clinical improvement during first-line treatment	Secondary	75% improved, 19% stable, 6% deteriorated; mean duration of improvement 3.4 months; no difference between arms
Radiologic improvement during first-line treatment	Secondary	54% improved, 31% stable, 14% progressed (no difference between arms, P=0.402); pseudoprogression in 49%
Overall grade 3–4 AE	Adverse	78%
Treatment-related deaths	Adverse	0
Eye AE (any grade) — more frequent with erlotinib	Adverse	P<0.0001
Skin AE (any grade) — more frequent with erlotinib	Adverse	P=0.004
Infectious AE (any grade) — more frequent with erlotinib	Adverse	P=0.042
Metabolic AE (any grade) — more frequent with everolimus	Adverse	P=0.0003
Severe (grade ≥3) skin AE — more frequent with erlotinib	Adverse	P<0.0001
Severe renal AE — more frequent with dasatinib	Adverse	P=0.0054
Severe gastrointestinal AE — more frequent with dasatinib	Adverse	P=0.038
Neurological severe AE	Adverse	No difference between arms (P=0.65); most frequent: headache, hydrocephalus, dizziness
Biopsy-related severe (grade 3–4) complications	Adverse	14 of 157 procedures reviewed; no biopsy-related deaths

Subgroup Analysis

Treatment effect homogeneous across age, sex, histone H3 type and biomarker-defined subgroups in the largest comparison (everolimus vs dasatinib). KEY THERANOSTIC FINDING: patients with PI3K/AKT/MTOR pathway mutations, mTOR activation gene-expression signature or chromosome 1q gain had median OS 14 months with everolimus vs 9 months with dasatinib. Four long-term survivors (>6 years) were all treated with mTOR inhibitor. KEY PROGNOSTIC FINDING: TP53 mutation — multivariate HR 2.84 (95% CI 1.92–4.20), P<0.0001 for OS; H3.1K27M favorable vs H3.3K27M (P=0.064 univariate; abolished after TP53 adjustment). Reirradiation rate (25% overall) similar across arms (P=0.32).

Criticisms

Trial designed >10 years ago when DIPG biology was less understood; biomarker selection (EGFR IHC, PTEN IHC) reflects 2014 knowledge.
First-generation kinase inhibitors used; newer-generation EGFR and mTOR inhibitors with better CNS penetration may yield different results.
No radiotherapy-only randomized control arm (deemed unethical by parent associations); historical-control comparison only.
Open-label design, with possible treatment crossover at progression (some patients switched drugs, 25% received reirradiation).
Smallest pairwise comparison (erlotinib vs everolimus) included only 59 patients — limited power for that arm.
Single coordinating center (Gustave Roussy) for historical controls; 72% of trial patients enrolled in France — generalizability outside Europe may be limited.
Theranostic mTOR benefit is exploratory and hypothesis-generating; requires prospective validation (planned in next-generation BIOMEDE adaptive trial vs ONC201).

Funding

Imagine for Margo, Étoile de Martin, Ligue Contre Le Cancer (Val de Marne and Haute-Savoie committees), Abbie's Army, CRIS Cancer Foundation, Cancer Research UK (DRCRPG-Nov21/100002 and C13468/A23536). Funders had no role in data analysis.

Based on: BIOMEDE (Nature Medicine, 2026)