Diffuse Intrinsic Pontine Gliomas (DIPG) appear almost exclusively in children and adolescents, representing 15 to 20% of posterior fossa tumours. Even if it is one of the most common malignant brain tumours, there are only 30 to 40 new cases per year in France. Their clinical presentation is stereotyped with a short clinical history and a unique MRI appearance that was usually considered as sufficient to establish the diagnosis. The prognosis of DIPG is always unfavourable; median overall survival is 9 to 10 months in general and most patients will die within two years after diagnosis (Kaplan 1996,Hargrave 2006). Malignant gliomas infiltrating the brainstem represent the greatest challenge of paediatric oncology; despite numerous collaborative studies performed, patients' survival has not significantly increased in thirty years (Hargrave 2009). There is no validated prognostic factor. There is currently no validated treatment except radiotherapy.

Several targeted agents have been tested in DIPG (Pollack 2007 Haas-Kogan 2008, Geoerger, 2011), without knowing whether the target was present in the tumour. A critical review of the paradigms of these trials tells us that there are long term survivors in these studies that is to say patients who may have benefited from the tested therapy, but they are few. So far, the new therapies that have been tried were evaluated one after the other in search of a treatment that would be effective for all patients, measuring the treatment effect on median survival. They were all rejected as ineffective. However the investigators can challenge the endpoint to evaluate efficacy in these trials as the existence of long term survivors (> 18 months, for example) and their number should not been ignored, especially if targeted therapies are considered. The investigators propose a paradigm shift in the choice of treatment; the issue raised would be to give to each patient the treatment associated with the highest likelihood of efficacy based on the specific biological tumour profile.

The development of targeted therapies for malignant gliomas infiltrating the brainstem has been hampered by the absence of biological data. It is therefore crucial to better understand the biology of these tumours. Despite the safety of the biopsy in brainstem tumours, most teams of paediatric neurosurgery limit the use of stereotactic biopsy only for clinically or radiologically unusual forms. Until recently, there has been no systematic genetic study at diagnosis to date and the few available data were confounded by the inclusion of autopsies or clinically and radiologically unusual cases (Louis, 1993; Gilbertson 2003; Okada, 2008; Zarghooni 2010; Broniscer, 2010; Wu, 2012 and Schwartzentruber, 2012).

French teams gathered in the French Society of Paediatric Oncology and the European consortium "Innovative Therapies in Children with Cancer (ITCC)" decided a few years ago to perform biopsies of these tumours for diagnostic confirmation and to ensure the presence of certain therapeutic targets prior to a possible inclusion in a trial evaluating a targeted therapy (Geoerger, 2009; Geoerger, 2010). Part of this experiment was reported by the team of the Necker Hospital in Paris, confirming the low rate of complications of stereotactic biopsy procedure (Roujeau, 2007). The biopsy specimen analysis allowed practicing immunohistochemical, genomic (CGHarray), gene expression (transcriptome) and direct sequencing of candidate genes studies.

In this study, the majority of patients will receive a treatment assumed to specifically target a biological abnormality identified on the biopsy. More importantly, patients will not receive a drug for which the identified target is absent.

In this first step of the protocol, the patients will thus be allocated to one of the three treatment groups as follows:

• If the tumor overexpresses EGFR without PTEN loss of expression, patients may receive erlotinib or dasatinib allocated by randomization (R1 randomisation).
• If the tumor shows loss of PTEN expression without EGFR overexpression, patients may receive everolimus or dasatinib allocated by randomisation (R2 randomisation).
• If the tumor shows both EGFR overexpression and loss of PTEN expression, patients may receive erlotinib, everolimus or dasatinib by randomisation (R3 randomisation).
• If the tumor shows neither EGFR overexpression nor loss of PTEN expression (a very rare situation in our experience), patients will receive dasatinib (no randomisation).
• If the biopsy assessment is not contributive, the treatment will be allocated by randomisation between erlotinib, everolimus and dasatinib (R3 randomisation).

Eligibility criteria for the BIOMEDE study (pre-screening for the randomised subtrials)

• Diagnosis of DIPG (clinical and radiological, or histological in case the biopsy was performed before study entry)
• DIPG at diagnosis: no prior chemotherapy for the present cancer;no prior cerebral radiation therapy
• NB : Metastatic disease allowed. Patient with metastatic disease are eligible for the study (including the randomised trial if diagnosis of DIPG confirmed). In this situation, radiotherapy will have to start within three weeks after the biopsy while targerted treatment will start at the end of the irradiation.
• Age > 6 months and < 25 years. For children below the age of 3 years, inclusion in the study and medical decisions should be discussed with the coordinating investigator.
• Eligible for a biopsy, or biopsy performed for diagnostic purpose and material available for the biomarker assessment
• Eligible for cerebral radiotherapy
• Patient covered by an health insurance if national requirement
• Written informed consent given by patient and/or parents/legal representative for biomarkers assessment and registration in the study.

Non eligibility criteria for the study

• Massive intratumour bleeding
• Any other concomitant anti-cancer treatment not foreseen by this protocol
• Any other cancer during the last 5 years
• Uncontrolled intercurrent illness or active infection
• Any other co-morbid condition that in the investigator's opinion would impair study participation
• Unable for medical follow-up (geographic, social or mental reasons)
• Patient not fulfilling one of the previous eligibility criteria.
• Patient previously treated with irradiation on the brainstem for another neoplasm
• Patient with congenital galactose intolerance, Lapp lactase deficiency or glucose-galactose malabsorption.
• Patient not covered by a social security agreement accepted in the treating country if national requirement
• Pregnant or breast feeding women
• NB: A patient with known hypersensitivity for one the drug or its excipients could still participate to the study and receive one of the other drug(s)

Common eligibility criteria for the BIOMEDE randomised subtrials

• Eligibility criteria for the study (see above)
• Confirmed histological diagnosis of diffuse intrinsic pontine glioma (grade II, III, IV WHO), confirmed by central pathology review (including the assessment of the loss of H3K27me3 by immunohistochemistry or the presence of a mutation in the histone H3 variant genes).

Patients without classical clinical and radiological diagnostic criteria who fulfil the histological and biological criteria of DIPG are eligible for the trial.

Pilocytic astrocytoma and gangliogliomas are not eligible.

• Life expectancy > 12 weeks after the start of study treatment
• Karnofsky performance status scale or Lansky Play Scale > 50%. The PS should not take the neurologic deficit per se into account. NB: Children and young adults with a worse performance status due to glioma-related motor paresis can be included.
• Absolute neutrophil count > 1.5 x 109/l, Platelets > 100 x 109/l
• Total bilirubin < 1,5 x ULN, AST and ALT< 2,5 x ULN
• Serum creatinine < 1,5 X ULN for age. If serum creatinine > 1,5 ULN, creatinine clearance must be > 70 ml/min/1,73 m² (EDTA radioisotope GFR or 24 hours urines collection)
• Normal coagulation tests: prothrombin rate (prothrombin time = PT), TCA (PTT), fibrinogen
• No current organ toxicity > grade 2 according to the NCI-CTCAE version 4.0 especially cardiovascular, pulmonary or renal disease (,including but not limited to: congenital long QT syndrome, nephrotic syndrome, glomerulopathy, uncontrolled high blood pressure despite adequate treatment, interstitial lung disease, pulmonary arterial hypertension). In case of known or possible cardiac disease, a cardiological advice will be required prior to the inclusion in the randomized trial as a preexisting cardiopathy represents a contra-indication to dasatinib.
• Effective contraception for patients (male and female) of reproductive potential during their entire participation in the study and during 6 months after the end of treatment
• Negative pregnancy test (serum beta-HCG) evaluated in the last week in females of reproductive potential
• Written informed consent given by patient and/or parents/legal representative for treatment and randomization

Eligibility criteria for the subtrials Eligibility criteria for the different subtrials will be mainly based on biomarkers assessment as detailed in the table above. In addition, contra-indication and precautions for use to specific drugs will be considered.