paediatrics Brussels 17

85

M. Reni et al. / Critical Reviews in Oncology/Hematology 63 (2007) 81–89

better evaluation of the degree of resection and may also iden- tify cases in whom immediate second-look surgery might be useful. Significant improvements in neurosurgical tech- niques and neuro-anesthesia have facilitated a reduction of operative mortality of 25–50% for infratentorial and of 6–22% for supratentorial ependymomas to less than 5% of cases [38,47,52,56] . Incomplete resection is the rule because ependymomas usually grow in highly specialised areas of the central nervous system. The frequency of complete resections is higher in surgical series than in radiotherapy series and ranges from 25% to 93% for supratentorial ependymomas [33,39,66,67] and from 5% to 72% for infratentorial ependy- momas [33,38,39,44,46,49,52,54,67] . The rate of gross total resections in infratentorial tumours depends on their location, being up to 100% in the roof of the fourth ventricle, 86% in mid-floor tumours and 54% in the lateral recesses [52] . Spinal cord tumours, which in the majority are low-grade lesions, can often completely be removed, and without functional sacrifice in 27–45% of cases [33,34,46,47,49,68–70] . Postoperative radiotherapy (RT) is the standard treatment on a type C basis in high-grade ependymomas, and on a type 3 basis in low-grade ependymomas. Five-year survival rates for patients with intracranial ependymomas have increased from 0–27% [71] to 36.5–87% since the use of systemic irra- diation [25,34,39,46,47,54–56,62,63,67,68,70,72–74] . How- ever, data on survival after surgery alone are limited inmodern series and a clear impact of radiotherapy on the outcome is not supported by consistent statistical data because no ran- domised trials have been carried out. The non-significant prolonged survival with postoperative irradiation as observed in retrospective series was either uncontrolled or compared with historical controls [45,54,72] while in the single series reporting a significant benefit, the comparison was biased due to inclusion of patients experiencing progressive dis- ease during postoperative chemotherapy, or patients younger than 3 years of age in the control arm [63] . Notwithstand- ing these considerations, particularly since failure to control local disease remains the most significant factor contribut- ing to recurrence and a poor survival, there is a consensus that radiotherapy should be part in the standard of care for the majority of patients. More recently, some small series reported a remarkably good outcome in children with low-grade intracranial ependymoma and did not receive irra- diation after a gross total resection [45,65–68,70,73,75,76] . The option of a close observation and delaying radiation until signs of tumour progression may be relevant, since late effects as cognitive deterioration, endocrine dysfunction in small children or dementia in the elderly constitute major concerns in patients who are potential long-term survivors of brain tumours. So, reserving RT for relapses appears to be an attractive strategy in these patients. It may also be considered as a therapeutic option for patients with a subtotal resection of low-grade ependymomas, assuming that the behaviour of 6.2. Radiation therapy

may have a somewhat better prognosis with 5-year survivals of 55–90% as compared to paediatric patients, i.e.14–60% [38,39,46,52,58] . In general, the younger the child the worse the prognosis [38,45,47,53] . It has been suggested that paedi- atric ependymomas may behave more aggressively, based on the more immature neural tissue of the children [38,58] . Dif- ferent patterns in cytogenetic aberrations between younger and older patients may underlie the age-related outcome [32] . The role of tumour location is also controversial. If spinal lesions are related to the most favourable outcome, things are less clear for intracranial tumours. Some authors reported no prognostic impact of this variable [37,39,49,54,55] . Accord- ing to others, supratentorial ependymomas are related to a worse prognosis because they more often exhibit periph- eral infiltrative growth into the brain parenchyma and are less often entirely encapsulated rendering surgical resectabil- ity troublesome [33,34,38,46,59] . Furthermore, infratentorial tumours show a lower mitotic activity than supratentorial tumours [60,61] . Conversely, other authors claimed a worse prognosis for ependymomas arising from the posterior fossa, which occur in younger patients [45] and invade the brain- stem, the floor of the fourth ventricle or cranial nerves of the cerebellopontine angle through the foramen of Luschka, precluding complete resection [38,53] . The 5-year survival ranges between 35% and 76% for supratentorial, 40% and 59% for infratentorial, and 57% and 100% for spinal ependy- momas [33,39,46,49,62] . The extent of resection has been proposed as an independent prognostic factor. Gross total resection achieves a better 5-year survival than a subtotal removal or biopsy [52,63,64] . In some cases, the benefit of complete resection was limited to low-grade tumours [55] . However, most authors failed to find any significant survival advantage related to the extent of resection [34,38,49,53,57] . The lack of evidence for the impact of surgery on survival could be related to the unreliability of subjective assess- ment of the degree of surgical ablation [65] . By contrast, the degree of resection, when assessed by postoperative imaging, revealed a significant difference in 5-year freedom from pro- gressive disease in a small series of 19 patients [65] . Female gender was reported to show better survival than male gender in one single series [55] .

6. Treatment

6.1. Surgery

Surgery represents the standard treatment for ependy- moma; it provides tissue for histologic diagnosis, may re-establish cerebrospinal fluid flow, and permits debulking or total resection of the tumour. Maximal surgical resec- tion can be of paramount importance and should be carried out whenever possible, without compromising neurological function, and a positive relation between extent of imaging- based surgical resection and outcome has been suggested [65] . Postoperative magnetic resonance imaging allows a