The initial surgical management is aiming at complete removal of the tumor, if necessary in more than one step because it remains the principal risk factor for survival. Indeed, in case of incomplete resection, one should not proceed to irradiation. Risk of relapse is five times higher in case of incomplete resection even after full-dose radiation therapy [1]. In the Saint-Jude series, presently the best results published so far, 5-year event-free survival is only 41 % (95 % CI = 18–64 %) in case of near-total or subtotal resection versus 81.5 % (95 CI = 73–90 %) in case of gross-total resection, p < 0.0001. Any attempt should therefore be made to obtain a macroscopically complete resection. In between two surgeries, there is no definitive indication for chemotherapy since most of the regimens tested have not shown many responses [2]. Pre-irradiation chemotherapy may compensate for an incomplete resection as it has been shown with a recently published phase II using a standard chemotherapy regimen based on cisplatin in patients with incompletely resected; in this report, patients with incompletely resected ependymomas treated with pre-irradiation chemotherapy had similar although slightly lower EFS than those with completely resected ependymomas treated with irradiation only [3]. In addition, some anecdotal reports have shown that second surgery may be facilitated by a course or two of chemotherapy by reducing vascularization and bleeding. Evaluating new agents could therefore be introduced as upfront window in between two surgeries; this strategy will be part of the next SIOP trial for ependymoma. Multiple steps are sometimes necessary to reach macroscopic resection before embarking into the adjuvant treatment.

In most of the cases, ependymomas cannot be cured by surgery only, and adjuvant treatment has to be associated. Surgery can however be enough to cure completely resected low-grade ependymomas in the spine or in the supratentorial brain. Small series of supratentorial low-grade ependymomas treated with surgery alone have shown that a small proportion of patients, around one third, may be cured without recourse to any adjuvant treatment [4, 5]. Actual guidelines for the management of myxopapillary and intramedullary ependymoma indicate that adjuvant treatment can be omitted after complete resection [6]. Although ependymomas occurring in the superior part of the spine tend to have a better prognosis, there is currently no consensus for a different management according to the location of the ependymoma in the spine [7].

In all the other situations, ependymomas should receive adjuvant treatment after surgery.

It has been the rule to avoid radiation therapy in young children in the past 20 years [8, 9]. Despite disappointing long-term results with only 30–50 % of relapse-free survivors at 5 years, this strategy allows to cure some patients without recourse to radiation therapy. It is, therefore, possible to choose this option either in children with very good prognostic factors, see below, or in children to young to receive irradiation. Presently and because of the progress of radiation techniques, irradiation can be considered in children down to 1 year for posterior fossa tumors where we know that the consequences of radiation therapy to the posterior fossa are limited [10, 11]. There is more debate for supratentorial ependymomas, but some studies have shown that cognition may be preserved with more conformal techniques [12]. Decision to embark with adjuvant radiation therapy in infants with supratentorial ependymomas may be defined on an individual basis considering both prognostic biomarkers and the size of the field of irradiation. For children that do not receive radiation therapy, dose dense chemotherapy such as the combinations used in the UK [9] or the intensified baby-POG regimen is a valuable option [13].

In the other children requiring adjuvant treatment, radiotherapy is delivered routinely with local fields because craniospinal irradiation was unable to prevent distant/metastatic relapses [14]. Recent studies have shown that the radiation field can be limited to the tumor bed without the need to irradiate the whole posterior fossa [9, 15]. The best results are obtained with higher doses of irradiation up to 59.4 Gy culminating in 5-year overall survival above 75 % [1]. These doses may have to be mitigated depending on the organ at risk such as the upper spinal cord or optic tracts.

There is currently no proof that the addition of chemotherapy on top of irradiation could improve the outcome of the patients. Indeed, the only randomized study published was underpowered and could not show any advantage with a relatively mild chemotherapy [16]. It will however be the purpose of the new SIOP ependymoma trial, to test the addition of a more intensive chemotherapy alternating cisplatin and the VEC (vincristine-etoposide-cyclophosphamide combination) [17].

Recurrences are mostly local despite the use of high dose of irradiation. After a chemotherapy-first approach, around 90 % of relapses are in the tumor bed [8, 9]. After a radiotherapy-first approach in older children however, the percentage of local relapses is slightly lower, i.e., around 60 % [1]. This calls for improvement in the local control, i.e., with an additional boost. The feasibility and safety of this approach will be tested as a phase I trial in incompletely resected tumors in the next SIOPE protocol.

Whenever possible, reoperation is key for disease control after relapse as shown irrespective of the initial treatment [8, 18]. Even when the relapse can be treated with re-irradiation, most of the patients whose disease can be durably controlled are those who could undergo a complete surgery of the relapse. This is true for local as well as distant relapses.

The benefit of chemotherapy is limited at this stage. Numerous chemotherapy regimens have been tested at relapse. Response rate for single agents are usually below 20 % except for cisplatin and etoposide [2]. Best results reported with combinations were seen with ICE [19] and VEC [17]. High-dose chemotherapy is of no benefit [20]. Antiangiogenic agents such as bevacizumab have also failed to control recurrent ependymomas [21]. Metronomic therapy has shown some benefit in selected cases [22]. Recent drug screen in preclinical models have shown that 5-FU may be an option [23]. New agents are definitively needed to improve the prognosis of ependymomas.

Re-irradiation has been attempted with many techniques depending on the type of relapse and on the availability of new radiation modalities: stereotactic hypofractionated irradiation [24, 25], conformal conventionally fractionated full-dose irradiation [18, 26], and Gamma knife [27]. Local relapses have been treated usually with local re-irradiation and metastatic relapses with craniospinal irradiation. Time to progression after re-irradiation has been shown to be possibly longer than the time to progression after the first irradiation [18, 24, 26]. Tolerance has been shown to be good in most of the cases, although radionecrosis is not uncommon [24]. Some patients could eventually receive more than one course of stereotactic re-irradiation. There has been no formal comparison of these different modalities. There is however a clear indication that these re-irradiations are effective in terms of tumor control provided that the recurrence is not metastatic, the tumor is supratentorial, and the interval since previous radiotherapy is superior to 18 months [25].

Grading of ependymomas is not reproducible and is rarely associated with survival as shown by an international panel of pathologists reviewing four different trials cohorts [28]. Numerous biomarkers have therefore been tested, usually only in single-center cohorts.

A small but significant proportion of young children with ependymomas can be cured with chemotherapy. Combining two trial cohorts of children below the age of 5, Kilday et al. could show that the 1q gain was an adverse prognostic factor for recurrence but not for survival [29]. Similar data in children treated with irradiation will soon be published by the German HIT group (Pietsch, personal communication). The value of this strong adverse biomarker is limited by the proportion of patients with this gain that is below 20 % and cannot account for all children who experience a relapse.

Ependymomas cannot be considered as a single entity anymore. Indeed, posterior fossa tumors have been divided into two groups according to the methylation profile of the DNA; a hypermethylated profile is associated with the expression of mesenchymal markers such as LAMA2 or tenascin-C and a worse survival [30, 31]. Epigenetic modifiers such as 5-aza-2’-deoxycytidine (a DNA-demethylating agent) or 3-deazaneplanocin A (an inhibitor of the methylation of histones) have an antitumor effect in vitro [31]. Likewise, supratentorial ependymomas are divided into a poor prognosis group based on the presence of a specific translocation involving the RELA gene [32] and a good prognostic group that could be characterized by the overexpression of neuronal markers [33]. These categories may be considered for treatment stratification, provided that they could be validated in independent cohorts.