Cancer in Children: Clinical Management, 5th Edition

Chapter 17. Tumours of the central nervous system

R. Pötter

T. Czech

K. Dieckmann

I. Slavc

D. Wimberger-Prayer

H. Budka


Tumours of the central nervous system (CNS) constitute the most common solid tumour in children (2.7 per 100 000 children annually), and therefore contribute to a major part of daily practice in paediatric oncology. Brain tumours are very heterogeneous with regard to tissue, location, pattern of spread, clinical picture, natural history, and age of occurrence (from the neonatal period to adolescence). As significant progress has been achieved and is developing in the different areas of diagnosis and treatment, adequate clinical management and follow-up now involves an interdisciplinary paediatric neuro-onocology team (neuropaediatrician, paediatric oncologist, neurosurgeon, radiotherapist, neuropathologist, neuroradiologist, and psychologist). Management decisions for childhood brain tumours should be based exclusively on such a team approach, in which the different members must be familiar with all the available knowledge, experience, and developments in their respective fields.1 The goal of brain tumour therapy is to achieve cure while avoiding unacceptable long-term sequelae.

Modern neurosurgery remains the mainstay of treatment for most brain tumours, followed by modern brain radiotherapy as the most important adjuvant procedure in a large number of patients.2 Over the past decade chemotherapy has been shown to be clearly beneficial for patients with medulloblastoma and high-grade glioma.3,4

Tumour classification and histologic diagnosis

Progress in our understanding of the molecular basis of neoplastic development may lead to molecular tumour diagnosis in the future. At present, however, descriptive classification by histologic examination remains pivotal for the appropriate management of CNS tumours. Since location and cellular differentiation are the basis of this diagnostic system, tumour classification has remained ‘histogenetic’. The most recent edition of the WHO Classification of tumours of the nervous system5 lists 127 entities, reflecting the remarkable variety of cellular constituents of the CNS. Theoretically, all types might develop in children. However, the number of tumour types which are of special importance in children is significantly smaller (Table 17.1). The majority of paediatric CNS tumours are of only five types: medulloblastoma, pilocytic astrocytoma, diffuse astrocytoma, ependymoma, and craniopharyngioma.

As with their counterparts in adults, CNS tumours in childhood have important properties which differ from those of tumours at other sites, but profoundly influence their behaviour:

·  many are highly invasive, even when histologically of low malignancy

·  many are heterogeneous in composition with areas of mixed tumour type and differing malignancy (special attention is necessary in stereotactic biopsies)


·  many are notorious for spreading to CSF pathways, thus enabling CSF seeding even in lowgrade tumours

·  tumours may progress from low to high grade, most prominently in diffuse astrocytomas.

Table 17.1. Histologic CNS tumour types of particular importance in children: most frequent age and site, WHO grade, and approximate percentage of all brain tumours in children

Tumour type

Most frequent site

Percentage of all brain tumoursa

WHO grade

Most frequent age

Pilocytic astrocytoma




Childhood and adolescence

Hypothalamus, optic pathways



Subependymal giant cell astrocytoma

Foramen of Munro region (usually in tuberous sclerosis)



Childhood, adolescence

Diffuse astrocytoma (low grade)

Cerebral hemispheres



Adolescence, young adults

Pleomorphic xanthoastrocytoma





Superficial cerebral hemispheres



Late childhood, adolescence

Anaplastic astrocytoma, glioblastoma

Cerebral hemispheres



All ages





Oligodendroglioma, mixed glioma

Cerebral hemisphere
Anaplastic oligodendroglioma



Adolescence, young adults

Ependymoma, anaplastic ependymoma

Lateral and third ventricles



Childhood and adolescence

Fourth ventricle



Choroid plexus papilloma, choroid plexus carcinoma

Lateral and fourth ventricles



Infancy, childhood

Gangliocytoma, ganglioglioma

Temporal lobe



Childhood and adolescence

Desmoplastic infantile ganglioglioma

Superficial cerebral hemispheres
Desmoplastic cerebral astrocytoma of infancy




Dysembryoplastic neuroepithelial tumour

Temporal lobe




Central neurocytoma

Lateral ventricles



Adolescence, young adults

Pineocytoma, pineoblastoma

Pineal region
Parenchymal tumour of intermediate differentiation


Childhood to young adults



Cerebral hemispheres



Infancy and childhood


Lateral and fourth ventricles



Infancy and childhood





Infancy and childhood

Other primitive neuroectodermal tumours

Whole neuroaxis



Infancy and childhood

Atypical teratoid/rhabdoid tumour

Infra-and supratentorial



Infancy and childhood

Germ cell tumours

Pineal region especially germinoma and teratoma



Infancy and childhood




Colloid cyst of third ventricle

Foramen of Munro region



Adolescence, young adults










All ages

Pituitary adenoma





Most CNS tumour types occur preferentially in specific age-groups and at specific sites (Table 17.1). Nevertheless, exceptions to these rules must be kept in mind. Compared with tumours in adults, the posterior fossa site is over-represented (about half of all paediatric CNS tumours occur at this site). However, the distribution depends upon age: during the first year of life and in adolescence, supratentorial tumours are more frequent than those in infratentorial sites, which predominate during childhood. Spinal cord tumours are rare in children. Many CNS tumours in the young age group occur near the midline, suggesting a developmental aspect to their origin.

Histologic grading of malignancy according to the WHO indicates how a given tumour type usually behaves on a four-point scale (Table 17.1). Again, exceptions may occur. In general, histologic malignancy is judged according to cellularity, pleomorphism, nuclear atypia, mitotic activity, invasiveness and metastasis, anaplasia (cellular differentiation), and secondary features such as necrosis and neovascularity. However, some of these items do not have the same meaning in all tumour types. It is important to separate pilocytic astrocytoma, which is the most frequent glioma of childhood and usually has an excellent prognosis, from the much less favourable diffuse astrocytomas.

The impact of modern morphologic techniques, with immunocytochemistry as an indispensable tool, on a refined histogenetic classification of CNS tumours cannot be overemphasized. Information on biologic tumour properties, such as proliferation, can also be obtained. While CNS tumour proliferation indices generally reflect the malignancy scale of the present classification system, proliferation on its own has not yet been shown to be of decisive prognostic importance. Thus it is the combined consideration of distinct features of the histologic examination (tumour type, grade of malignancy, growth pattern, proliferative activity, etc.), which will provide the maximum relevant information for further management and therapy of an individual tumour patient.

The concept of the primitive neuroectodermal tumour

In Bailey and Cushing's original histogenetic tumour classification, tumour morphology was considered to mirror specific stages of normal neural tissue development. Since then, highly undifferentiated neural neoplasms have been considered as ‘embryonal’, with the medulloblastoma as the most important clinicpathologic entity. More recently, the unifying concept of the primitive neuroectodermal tumour (PNET) was proposed to encompass all types of undifferentiated or primitive small-celled neural neoplasms with a potential for multiple differentiation (neuronal, glial, ependymal, pineal, etc.), irrespective of their site of origin. The PNET concept has become generally accepted and has entered the present WHO classification of embryonal tumours in a somewhat hybrid format. The distinct clinicopathologic tumour entities with traditional terminology such as the medulloepithelioma, medulloblastoma, and ependymoblastoma are retained in addition to PNETs in supratentorial presentations. Thus PNET may be used as a general term for all types of densely cellular (‘blue-celled’), primitive, or ‘embryonal’ neural tumours. Whenever possible, however, further delineation as medulloblastoma etc. should be given. Confusingly, ‘PNET’ (peripheral neuroectodermal tumour) is sometimes used to designate a peripheral nervous system tumour with similarities to Ewing sarcoma.

‘New’ CNS tumour entities of childhood

In addition to earlier accepted PNETs and the long list of traditional tumour entities, the more recent editions of the WHO classification (1997 and 2000) include a number of recently recognized tumour entities, including some that usually present in childhood.5 The now well-established pleomorphic xanthoastrocytoma is a predominantly extracerebral growth with pleomorphic, lipidized astrocytes and a generally, but not universally, favourable prognosis. The desmoplastic infantile ganglioglioma and its variant, the desmoplastic cerebral astrocytoma of infancy, are usually very large tumours presenting during the first 2 years of life. The dysembryoplastic neuroepithelial tumour usually presents in a temporal location with long-standing complex partial seizures; it features a mixed glioneuronal population in a mucinous matrix, distributed in characteristic multiple cortical nodules. The central neurocytoma is almost invariably an intraventricular tumour of the lateral ventricle, and has been misdiagnosed for decades as oligodendroglioma or ependymoma; its prognosis is mostly, but not always, favourable. The pineal parenchymal tumour of intermediate differentiation has a variable clinical behaviour and constitutes about 10 per cent of all pineal parenchymal tumours. An important new addition to the group of embryonal tumours is the atypical teratoid/rhabdoid tumour which constitutes 2.1 per cent of a group of primary CNS tumours of children. It manifests more frequently infratentorially than supratentorially with a histology of rhabdoid cells (large eosinophilic cell bodies with eccentric nucleus), with or without PNETlike areas, epithelial or mesenchymal structures, and a complex immunocytochemical expression of glial, neuronal, epithelial, mesenchymal, and muscle proteins. Most patients die within a year of diagnosis.

Molecular biology

Over the past two decades, work in the field of molecular neuro-oncology has evolved from being largely descriptive in nature to the implication of specific genes and the functional analysis of their role in the evolution of the malignant process. Cytogenetic and molecular techniques have resulted in the cloning of the retinoblastoma, the neurofibromatosis type 1 and type 2 genes, and the characterization of the molecular physiology of the TP53 gene as well as genes involved in the inherited genetic disorders discussed below. These efforts will help to improve understanding of the many and complex pathways that lead to tumour initiation and progression, and to guide therapeutic strategies.

The most common and sometimes the only structural abnormality seen in medulloblastoma is an isochromosome 17q [i(17q)] present in approximately 50 per cent of cases. This finding implicates the presence of a tumour suppressor gene on 17p, which is important in tumour development. So far, a number of genes on 17p have been eliminated as candidates for this locus, including TP53.

Recently, two independent groups in Boston and Philadelphia found in retrospective studies of medulloblastomas that high neurotrophin receptor TrkC mRNA expression is a powerful independent predictor of a favourable clinical outcome in medulloblastoma patients.7,8 In addition, independent studies in Göttingen and Philadelphia identified low proto-oncogene MYC (c-myc) mRNA expression as a second independent predictor of a favourable survival outcome.9,10 MYC mRNA expression was found to range widely and did not correlate with the presence of MYC gene amplification in medulloblastoma cell lines or primary tumours.

Other biologic prognostic indicators include increased expression levels of the neuregulin receptors ErbB2 and ErbB4,11 which were found to be highly correlated with poor outcome.

Pomeroy et al12 recently extended these findings by demonstrating that predictors based on microarray gene expression profiles can predict clinical outcome with high statistical significance. Using oligonucleotide microarray-based gene expression profiling to monitor the expression of over 6800 genes in the tumours of children with medulloblastoma, 5–20 marker gene predictors were found to be accurate outcome predictors, performing significantly better than clinical staging.

Other studies have examined the functional role of genes implicated in glioma malignancy and investigated the mechanisms of their dysregulation in the generation of the malignant phenotype. These studies suggest that there are at least two pathways that lead to glioblastoma in adults. The first pathway involves progression from a diffuse astrocytoma to an anaplastic astrocytoma to a glioblastoma multiforme. An early event in this progression is loss of the short arm of chromosome 17 in combination with an inactivating mutation in the retained copy of TP53. Later events in this pathway may involve alterations of genes on chromosome 9, 19, and 22. The second pathway suggests the formation of a de novo glioblastoma, rather than progression from a lower-grade tumour, and involves loss of heterozygosity on chromosome 10. Chromosome 10 loss, but not 17p loss, is often associated with amplification and rearrangement of the epidermal growth factor receptor gene, resulting in increased cell proliferation.

Interestingly, work on paediatric astrocytomas suggests that the genes involved are different. In contrast with glioblastoma multiforme in adults, paediatric glioblastomas generally lack EGFRMDM2 and/or CDK4 gene amplification, and chromosome 10 deletions are rare.13 Taken together, these observations imply that the pathways leading to the development of malignant astrocytomas in children may differ significantly from those involved in adults, which may in part account for the somewhat better prognosis of such lesions in children. The only paediatric tumour that exhibits many of the same chromosomal abnormalities as the adult tumours is brainstem glioma, perhaps explaining its dismal prognosis.14


CNS tumours represent the second most frequent malignancy and the most common type of solid tumour in children <16 years of age. The annual incidence rate for all CNS tumours combined is approximately 2.5 cases per 100 000 children annually, with astroglial tumours accounting for 60.9 per cent and neuroectodermal tumours for 23.9 per cent.15 Approximately 8 per cent of these children are ≤2 years of age at the time of diagnosis.16

There are still no identifiable predisposing or aetiologic factors for the majority of children with brain tumours. However, in a small group of children the occurrence of primary CNS tumours is associated with certain hereditary and congenital diseases such as the neurocutaneous syndromes and the Li–Fraumeni syndrome. The best known of these is neurofibromatosis type I, in which there is a high incidence of visual pathway gliomas and other glial tumours. The hallmark of neurofibromatosis type II is the development of bilateral vestibular Schwannomas. In tuberous sclerosis, subependymal giant cell and other glial tumours may develop. In patients with von Hippel–Lindau syndrome, cerebellar haemangioblastoma, pheochromocytoma, renal cell carcinoma, and retinal tumours have been observed. Medulloblastoma and other malignancies may occur in association with the autosomal dominant nevoid basal cell carcinoma syndrome (Gorlin–Goltz syndrome) as well as Turcot's syndrome and ataxia–telangectasia. The most common CNS tumour in individuals with Li–Fraumeni syndrom is astrocytoma. Pineoblastomas can be seen in some cases of familial bilateral retinoblastoma.

Another confirmed aetiology for brain tumours in children is exposure to ionizing therapeutic radiation.

Clinical presentation

The signs and symptoms of neurologic dysfunction in a child with a brain tumour are various and depend more on age, premorbid developmental level, and site of origin than history. Brain tumours may cause neurologic impairment directly, by infiltrating or compressing normal CNS structures, or indirectly, by causing obstruction of cerebrospinal fluid (CSF) flow and increased intracranial pressure (ICP). The latter is responsible for the ‘classic’ triad of ICP— morning headache, vomiting, and visual disturbances. When present, these symptoms strongly suggest a rapidly growing midline or posterior fossa tumour. More commonly, the initial signs of ICP are more subtle, subacute, non-specific, and non-localizing. In school-age children, slowly developing ICP may be accompanied by declining academic performance, fatigue, changes in affect, energy level, motivation, personality, and behaviour, and complaints of vague intermittent headaches. In the first few years of life, irritability, anorexia, and developmental delay, with later regression of intellectual and motor abilities, are frequently early signs of ICP.

Infratentorial tumours (brainstem and cerebellar) commonly present with deficits of balance or brainstem function (truncal steadiness, extremity coordination and gait, cranial nerve function). Nystagmus and gaze palsy alone, or more likely in combination with deficits of cranial nerves V, VII, and IX, strongly suggest invasion of the brainstem. Head tilt may be a presenting sign.

Supratentorial tumours may cause a variety of signs and symptoms, depending on the size and location of the tumour. The most common presenting complaint is headache, followed secondly by seizures. Upper motor neuron signs such as hemiparesis, hyper-reflexia, and clonus, as well as associated sensory loss, may also be present.

Anorexia, bulimia, weight loss or gain, somnolence, mood swings, failure to thrive, diabetes insipidus, sexual precocity, or delayed puberty may be non-specific or suggest hypothalamic or pituitary dysfunction. Tumours of the region of the hypothalamus may also cause visual loss due to compression of the optic chiasm or the optic nerve.

Diagnosis and investigation

The advent of MRI has greatly altered the preand postoperative evaluation of children with brain tumours. Multiplanar imaging is extremely helpful in determining the exact extent of the tumour and its relationship to surrounding normal structures. The importance of CT is restricted to the proof of calcifications or associated bone destruction. The administration of paramagnetic resonance contrast agents, such as gadolinium diethylenetriaminepentaacetic acid (DTPA) in MRI studies and iodinated contrast agents with CT, have been proved to identify tumours and metastatic disease in brain and spine more accurately.17 In addition to MRI, magnetic resonance spectroscopy (MRS) has been shown to be helpful in differential diagnosis, management, and prognostication of brain tumours (Fig. 17.1). Positron emission tomography (PET) and single photon emission computed tomography (SPECT) are functional imaging tools which are increasingly used.

Sonography is the appropriate first investigation in children with open fontanelles. The children do not have to be sedated or transferred to an imaging center, sections can be obtained in a variety of orientations, and examinations can be repeated frequently. However, if a tumour is diagnosed by ultrasound, MRI or, if this is not available, CT is required for additional information.

Fig. 17.1 Magnetic resonance spectra of a brainstem tumour and a cerebral glioma. The distribution and intensity of the different substances [choline (Cho), creatine (Crea), and acetyl aspartate (Naa)] are characteristic of the malignancy of the tissue. The pattern of the brainstem lesion indicates high malignancy; that of the cerebral glioma indicates low malignancy.

Angiography is usually not needed unless a highly vascular lesion is suspected or a vascular malformation cannot be ruled out by non-invasive studies. In most cases, MR angiography, combined with a perfusion study, is sufficient to answer these questions.

Spinal imaging

Leptomeningeal metastases may affect the entire neuroaxis and necessitate evaluation of both the brain and the spinal cord. MRI has become the imaging method of choice as it has been shown to be clearly superior to myelography or CT myelography. When a brain tumour which is particularly prone to spinal seeding is diagnosed, MRI of the spine must be done before surgery of the primary tumour; otherwise postoperative reactive meningeal enhancement may mimic tumour involvement.

Postoperative imaging assessment

Surveillance imaging of the CNS at predetermined intervals is used to document the extent of tumour resection, to assess the tumour response to adjuvant treatment, to detect recurrence, and to evaluate the complications of treatment. The first contrast-enhanced postoperative MRI or CT scan should be scheduled within 72 h of surgery; within this period, enhancing structures are due to residual disease and not to therapy-related blood–brain barrier disruption. Consecutive scans can then be reliably compared with this ‘baseline’study. However, distinction between residual tumour and post-therapeutic effects can also be made by means of MRS.

Treatment modalities

Neurosurgical treatment

Surgical intervention remains the mainstay of the diagnostic and therapeutic management of primary brain tumours. The goals of the surgery are:

·  to establish a tissue diagnosis

·  to excise or reduce the tumour volume and the mass effect

·  potentially to cure the disease.

Completeness of surgical excision is the most important prognostic factor. Accurate histologic classification of a tumour is crucial for planning further therapy and determining prognosis. MRIand CT-guided stereotactic biopsy is used as a first step in selected cases of thalamic and basal ganglia tumours, multifocal lesions, and radiologically diffusely infiltrating lesions without significant mass effect.

Most paediatric neurosurgeons prefer open craniotomy to stereotactic biopsy in the majority of children, even when only a subtotal resection is anticipated preoperatively. As the expanding mass or resulting hydrocephalus leads to raised ICP, with the risk of herniation through either the tentorial notch or the foramen magnum, prompt and effective cytoreduction can be lifesaving, even when the tumour is only subtotally resected.

After acquisition of diagnostic imaging results, the child is prepared for surgery by first managing the most urgent symptomatic problems. This includes steroid administration to relieve accompanying brain oedema, and anticonvulsive treatment if indicated in supratentorial hemispheric tumours. In patients who present with hydrocephalus, a CSF shunt should be avoided. Endoscopic third ventriculostomy is the procedure of choice in obstructing hydrocephalus. If removal of the mass is anticipated, the decision about third ventriculostomy can be made after surgery, as permanent hydrocephalus occurs in only 25–30 per cent of all patients with posterior fossa tumours and hydrocephalus.

The safety and efficacy of surgery has improved because of advances in surgical techniques, progress in neuroimaging, and developments in neuroanesthesia and paediatric critical care.

Consideration of the anatomical location of the tumour as visualized by MRI [Fig. 17.2(a)], the knowledge of the possibilities for postoperative adjuvant therapy depending on tumour type (determined by frozen section and smear), the age of the patient, and ultimately the intraoperative findings will determine surgical aggressiveness. Surgery alone is increasingly able to cure some paediatric brain tumours or to keep them in extended remission. Tumours that invade deep grey nuclei, such as the hypothalamus, thalamus, or basal ganglia in the dominant hemisphere, tumours in eloquent cortical areas, and tumours that are intrinsic to the brainstem can often not be totally resected without significant risk of devastating neurologic sequelae. Technical adjuncts allowing safer and more complete resection of radiologically welldefined tumours include intraoperative ultrasound, frameless stereotactic techniques allowing interactive 3D-image-guided procedures [Fig.17.2(b)], and neurophysiologic monitoring of cortical, subcortical, and cranial nerve function.18 The Cavitron ultrasonic aspirator (CUSA) enables the removal of firm tumours while minimizing injury to surrounding structures. The surgical laser is advocated by some as helpful, especially in dealing with intrinsic spinal cord tumours.

Fig. 17.2 (a) MRI scan of a low-grade glioma in the occipital lobe. (b) The exposed cortex over the tumour with the contour of the lesion projected onto the operative site through a stereotactic surgical microscope.

Endoscopy may also assist the microsurgical procedure in evaluating the completeness of a resection, and offer alternatives for biopsy, or even resection, of intraventricular masses.


Radiotherapy has been known to be an effective modality in the treatment of childhood brain tumours for decades. Nevertheless, local control and survival, in particular in aggressive and invasive tumours, still need to be significantly improved. On the other hand, late side effects related to radiotherapy have been recognized more clearly with the increase in numbers and follow-up of long-term survivors.19,20

Current treatment policies, including radiotherapy, therefore aim to increase the therapeutic ratio by different means: minimizing side effects by limiting the indications for radiotherapy for good-prognosis tumours (e.g. low-grade astrocytomas accessible by modern neurosurgical techniques), postponing or avoiding radiotherapy in very young children (e.g. ependymoma, medulloblastoma), dose reduction in radiosensitive tumours (e.g. craniospinal irradiation in germ cell tumours), volume reduction through better adaptation of the treated volume to the target (e.g. circumscribed low-grade tumours), improving local control by improving the quality of radiotherapy (e.g. medulloblastoma), or dose escalation using modern radiotherapy techniques (e.g. stereotactic irradiation, intensity-modulated radiotherapy) in the case of less radiosensitive tumours (e.g. high-grade glioma, ependymoma).

Radiation-related side effects in CNS radiotherapy are generally related to total dose, fractionation of dose, radiation volume, area of CNS irradiated, and the age of the child at the time of treatment.

With advances in neurologic and biologic imaging (MRI, MRS, SPECT, PET), target definition for radiotherapy has improved or will improve considerably in the future in nearly all CNS tumours. With the integration of sectional, functional, and biologic imaging tools into 3D treatment-planning procedures and ‘conformal radiotherapy’ techniques (using multiple individually shaped fields), the treated volume can be more adequately tailored to the target volume. Thus, in conformal radiotherapy, the high-dose irradiation of normal brain tissue can be reduced, without jeopardizing treatment results, provided that target definition has been correct. Conformal radiotherapy techniques based on 3D treatment planning have become general clinical practice in the majority of advanced radiotherapy departments in recent years, in particular for brain tumours.

A precondition for such precision radiotherapy is immobilization of the patient, which only permits minimal variations throughout the planning procedure and the whole course of fractionated radiotherapy. A variety of custom-made immobilization devices are available and should be in common use for the treatment of childhood brain tumours.

One of the most difficult procedures is craniospinal irradiation (CSI), in which children are treated in a prone position using an individually adapted fixation system (e.g. a plaster immobilization device) for the head and the spinal axis. Detailed quality assurance procedures are necessary and should be performed at regular intervals (e.g. weekly) during the treatment course to check the accuracy of the treatment set-up.

Major technical advances in dose-delivery systems have been the introduction of stereotactic radiotherapy and intensity-modulated radiotherapy. The advantage of these techniques, which use multiple radiation fields or arcs, is the steep fall-off of radiation dose in all directions outside the treated volume. By using this technique, small tumours (up to ~4 cm in diameter) directly adjacent to critical structures can be treated with high radiation doses with reduced radiation-related morbidity. In future, stereotactic irradiation and intensity-modulated radiotherapy, in combination with biologic imaging, may allow a highly focused dose escalation with better local control. The efficacy of 3D conformal proton radiotherapy with its very steep falloff in dose (Bragg peak) has been demonstrated for low-grade tumours (e.g. large optic pathway tumours) in some specialized centers. With growing availability of proton treatment facilities, this most conformal radiotherapy will be increasingly used in paediatric brain tumours as well.

Image-guided stereotactic low-dose-rate interstitial and intracavitary brain radiotherapy with different isotopes (e.g. 125I, 192Ir 90Y, 32P) has been in use in some specialist centers and may be indicated in specific situations in slowly proliferating well-circumscribed lesions. However, great experience is necessary to handle these techniques adequately, and as the results reported so far are not clearly superior, they should only be used in experienced hands and in prospective clinical protocols.


Progress in the chemotherapy of brain tumours has lagged behind that of other neoplasms of childhood. Reasons for this include problems unique to the treatment of tumours of the CNS.

Special aspects of brain tumours

One cause of ineffectiveness of certain chemotherapeutic agents is the blood–brain/CSF barrier. While the role of this barrier is questionable in some tumours, such as medulloblastoma, it appears to be a more important issue in diffuse infiltrating gliomas as it is often difficult to deliver adequate concentrations of drugs to the periphery of the tumour. In addition, many brain tumours have a low mitotic index and thus a reduced susceptibility to cell cycle specific drugs. Another obstacle is brain tumour heterogeneity. Each entity has its own biologic characteristics, including cellular mechanisms within the tumour that are responsible for the ability to retain a chemotherapeutic agent or to repair drug-induced damage. A further problem is the tendency of paediatric malignant brain tumours to spread throughout the nervous system early during the course of disease. Therefore cytocidal drug levels must be obtained in the tumour tissue, brain parenchyma, and CSF.

Role of chemotherapy in brain tumours

Over the past decade chemotherapy has been shown to be clearly beneficial for medulloblastoma and high-grade glioma, and patients who received chemotherapy in addition to surgery and radiotherapy experienced better survival rates.

However, in paediatric neuro-oncology the use of chemotherapy is not limited to trying to improve survival. Given the long-term toxicities associated with whole-brain and/or neuraxis irradiation in children, chemotherapy is also used to delay and possibly replace some of the radiotherapy otherwise needed. In fact, in some patients, the need for radiotherapy has been eliminated by the use of chemotherapy. Chemotherapy is now widely used for infants and young children with malignant brain tumours and for patients with low-grade tumours that are not surgically resectable.

Chemotherapeutic agents and drug-delivery strategies utilized in brain tumours

Significant response rates have been reported for a number of chemotherapeutic agents. These include vincristine, procarbazine, the nitrosoureas BCNU and CCNU, etoposide, cisplatin, carboplatin, cyclophosphamide, ifosfamide, methotrexate, and temozolomide. As for most childhood cancers, multi-agent chemotherapy is utilized in brain tumours to improve efficacy and reduce the likelihood of resistance. Another strategy to improve the response rate of poorprognosis CNS tumours is high-dose chemotherapy with autologous stem cell rescue. Early reports in both newly diagnosed and recurrent CNS tumours have shown some encouraging results for subgroups of patients, and new trials have begun in several centers.

Strategies to overcome problems posed by the blood–brain/CSF barrier

As many brain tumours will spread throughout the CSF, future design of brain tumour protocols may include administration of a drug directly into the spinal fluid. The intrathecal route offers significant therapeutic advantage for molecules too large to pass through the blood–brain barrier and may avoid systemic toxicity.

The steep dose–response curve of bifunctional alkylating agents, coupled with their known activity against a variety of paediatric brain tumours, suggest the need for strategies to achieve high levels of these drugs in the CSF. Recent trials have demonstrated that agents such as mafosfamide, a preactivated cyclophosphamide derivative, can be safely administered into the CSF and may produce responses in leptomeningeal neoplasia. Intratumoral administration of chemotherapeutic agents is also undergoing investigation as a new modality to improve drug delivery to the tumour.

Since it does not appear that surgery or radiotherapy alone will improve the cure rate of brain tumours in the near future, the search must continue to identify new and better agents and to develop more effective delivery mechanisms for these agents. It is hoped that eventually better understanding of the molecular genetics and biology of brain tumours will translate into new treatments, including immunotherapy, gene therapy, and the use of anti-angiogenesis factors and second-messenger inhibitors.

Specific management

Low-grade gliomas

Low-grade gliomas account for 40 per cent of all childhood brain tumours and comprise a heterogenous group with regard to histologic subtype, anatomical location, and biologic behaviour.21 Childhood cancer registries assume a systematic under-reporting of these neoplasms which is partially due to the limited patient referral to centers of tertiary care.22 Although most children with low-grade gliomas already have an excellent prognosis, a proportion of these children will have recurrences following resection or experience progression following incomplete tumour removal or biopsy. Therefore the International Consortium of Low-Grade Glioma Research developed a protocol (SIOP–LGG 2003) which offers a comprehensive treatment strategy for all children and adolescents who are affected by a low-grade glioma arising in any part of the CNS. Results of the preceding SIOP–LGG trial, as well as from national trials and reports in the literature, form the basis of the recommendations and the randomized part(s) of the study.

Cerebellar astrocytomas

Cerebellar astrocytomas carry a more favourable prognosis than most other brain tumours.23 The majority are histologically benign, slow-growing, well-circumscribed, and often cystic lesions which involve the vermis and cerebellar hemispheres with approximately equal frequency. Invasion of the cerebellar peduncles or brainstem may occur. The goal in the treatment of these tumours is gross total resection. If achieved, a cure rate of nearly 100 per cent is expected. However, in some children, total removal may be impossible or hazardous due to brainstem involvement or perioperative complications. Although subtotal resection may allow an extended period of disease control in these patients, a significant percentage of lesions ultimately progress and require additional therapy. At present, the role of radiotherapy after incomplete resection remains unclear. In the absence of convincing data favouring the routine use of radiotherapy, many groups defer it until there is evidence of progressive disease that is surgically unresectable. Early experience at some institutions with radiosurgery and stereotactic radiotherapy for the treatment of focal areas of tumour recurrence suggests that this modality is useful in managing small areas of unresectable disease in critical locations.

Gliomas of the optic pathway

Most gliomas of the optic pathways occur during the first 5 years of life and are low-grade pilocytic astrocytoma. These tumours are associated with neurofibromatosis (NF-1) in 15–20 per cent of patients. Clinical symptoms depend on the location along the optic pathway, and include exophthalmos, decrease in visual acuity, disk pallor, visual field changes, endocrine dysfunction, and the diencephalic syndrome. The behaviour of these tumours is highly variable and unpredictable. Large tumours may remain stable for years, while initially small chiasmatic tumours may show rapid disease progression.

Management strategies for visual pathway gliomas include observation, chemotherapy, radiotherapy, surgery, and various combinations of these modalities. The time for intervention and choice of treatment modality have to be considered carefully. The risks of treatment-related side effects, such as optic nerve injury, vasculopathy, and endocrinopathy, have to be weighed against irreversible symptoms of tumour progression, leading to deterioration of visual and endocrine function.

Children without evidence of tumour growth after diagnosis are followed by MRI and neurodevelopmental, ophthalmologic, and endocrinologic examinations. When tumour progression is diagnosed, therapeutic intervention should be considered.

Unilateral optic nerve gliomas without chiasmal involvement may be cured by resection of the affected optic nerve. An alternative approach which conserves the optic nerve is conformal stereotactic radiotherapy. Tumours involving the chiasm and hypothalamic tumours are treated by a combined approach including surgery, radiotherapy, and chemotherapy.

Neurosurgical intervention aims at tissue diagnosis, resection of exophytic tumour extensions, drainage of tumour cysts, and decompression of the optic nerve.

The role of chemotherapy for the treatment of young children with progressive symptomatic tumours can be considered fairly well established in terms of achieving tumour responses and prolonged progression-free survival, and delaying the need for radiotherapy.24,25,26 According to SIOP–LGG 2003, children ≥8 years of age with visual pathway tumours and without NF-1 will be randomized to receive standard induction chemotherapy with vincristine and carboplatin or intensified induction chemotherapy with vincristine, carboplatin and etoposide. In older children (8 years) radiotherapy is the preferred first adjuvant therapy. Children affected by NF-1 are particularly prone to developing vasculopathies and should be treated with chemotherapy at all ages.

Local radiotherapy is performed by applying radiation doses of 40–50 Gy with highly focused conformal and stereotactic techniques to limit irradiation of uninvolved tissues, where possible. If there is tumour progression during or after chemotherapy, local radiotherapy remains the treatment of choice.

Low-grade brainstem gliomas

Low-grade tumours of the brainstem are typically ‘focal’ and can be differentiated from the prognostically unfavourable diffuse tumours by MRI. They make up ~20 per cent of tumours in this location and can be further classified on the basis of their topography and growth pattern. Clinical symptoms depend on location and tumour size, and often consist of cranial nerve deficits, long tract signs, and ataxia, whereas symptoms of hydrocephalus are present in the dorsally exophytic tumours of the pons and medulla filling the fourth ventricle and are the typical presentation of tectal tumours. This subgroup of midbrain tumours of the quadrigeminal plate with obstructive hydrocephalus usually has an indolent course, and treating the hydrocephalus with an endoscopic third ventriculostomy without histologic confirmation is considered appropriate. The histology of the focal tumours of the brainstem is mostly that of a low-grade lesion and includes astrocytoma grade 2, pilocytic astroytoma grade 1, and ganglioglioma. Rapidly progressing symptoms in a patient with a focal tumour on MRI are suggestive of a high-grade tumour (e.g. PNET).

Surgical resection should be discussed for focal low-grade tumours which can be reached without unacceptable morbidity. While complete resection is possible in selected cases, the residual disease may remain stable without need for adjuvant therapy even with partial resection.27 In the case of progression or non-resectability with severe symptoms, chemotherapy should be considered according to current trials. An interesting alternative is (stereotactic) conformal radiotherapy which is being investigated in some centers.

Supratentorial low-grade gliomas

Low-grade gliomas comprise a heterogeneous group with regard to histologic subtype, anatomical location, and biologic behaviour.

Roughly half of supratentorial low-grade gliomas are located in the cerebral hemispheres, and the remainder occur in the deep midline structures of the diencephalon and basal ganglia. Pilocytic and diffuse astrocytomas are the most frequently encountered gliomas, although other variants, such as the ganglioglioma, pleomorphic xanthoastrocytoma, and oligodendroglioma, must also be considered in the differential diagnosis. Surgical excision remains the primary therapy for the majority of these low-grade gliomas. Since at least 50 per cent of children with low-grade gliomas of the cerebral hemispheres present with seizures, the goal of surgery includes the alleviation of an associated seizure disorder, when intractable. It is now possible to target the operative approach to a subcortical lesion or a superficial lesion that is located within ‘eloquent’ cortex by using a combination of functional studies and stereotactic localization. Provided that a total excision can be achieved, no further therapy is warranted. Conservative management with routine imaging follow-up is appropriate for those lesions that are incompletely resected as childhood tumours rarely progress histologically to more malignant lesions. Reoperation is necessary if recurrence is documented, and radiotherapy is utilized for those lesions that are incompletely resected following recurrence.18

Unitl recently, thalamic astrocytomas were considered to be largely unresectable. However, with the implementation of computer-assisted stereotactic approaches, perioperative morbidity and mortality have dropped substantially, and near-complete resection has become an attainable goal in many children with pilocytic, low-grade, and cystic astrocytomas. In such cases, adjuvant therapy can often be deferred. For patients in whom a subtotal resection is performed to avoid the risk of incurring neurologic deficit, long-term disease-free survival can occur with certain indolent low-grade astrocytomas which correspond histologically to pilocytic astrocytoma of the cerebellum or optic pathway. Adjuvant therapy, either radiation (e.g. stereotactic procedures) or chemotherapy, is utilized in those cases of low-grade lesions that are unresectable and have documented disease progression.

High-grade gliomas

Anaplastic astrocytoma and glioblastoma

Most malignant gliomas, other than brainstem gliomas, are supratentorial in location and are among the most difficult tumours to treat in children. With a combination of surgery and irradiation, the median survival for children with malignant gliomas is only 9 months. In a randomized Children's Cancer Support Group trial (1985–1990), a multidrug regimen (‘eight in one’) was tested against CCNU, vincristine, and prednisone which had been evaluated in the first trial (1976–1981). No difference was detected. However, both groups had outcomes superior to standard irradiation and surgery. Five-year-survival for anaplastic astrocytoma and glioblastoma was 42 per cent and 27 per cent, respectively, for patients with >90 per cent resection, compared with only 14 per cent and 4 per cent, respectively, with less resection. However, it remains uncertain whether this survival advantage is a direct result of surgery, or merely reflects the fact that certain tumours, by virtue of their less aggressive growth characteristics or more favourable location, are amenable to more extensive resection.28

In Germany children with high-grade gliomas are treated according to recently proposed prospective cohort studies (HIT-GBM-D 2003). All patients receive simultaneous radiochemotherapy consisting of external beam therapy (54–59 Gy) and cisplatin, ifosfamide, etoposide, and vincristine followed by maintenance therapy with CCNU, vincristine, and prednisone, a combination which has shown efficacy in the earlier trials mentioned above. In one of the two treatment arms patients receive high-dose methotrexate before radiotherapy. Survival rates of the prior HIT GBM trials (A, B, C) for patients with completely resected supratentorial high-grade gliomas were 48 per cent after 5 years. For patients in whom the tumour could not be resected the 3-year survival was only 3 per cent (J. Wolff et al., unpublished data).

High-grade brainstem glioma

Approximately 80 per cent of tumours arising in the brainstem are diffuse intrinsic lesions which mostly carry an unfavourable prognosis. They can be distinguished from the favourable glioma by their typical diffuse growth pattern as seen on MRI. As a rule, there is no need for histologic confirmation by stereotactic biopsy if the clinical presentation and the MRI findings are characteristic, even though stereotactic procedures now carry a low risk of morbidity.

Clinical symptoms often consist of deficits in cranial nerves V–IX, long tract signs, and ataxia, and depend on location and tumour size. A short duration of symptoms (e.g. 1 month) and diffuse infiltration on MRI indicate a high grade of malignancy and an unfavourable prognosis.

Radiotherapy remains the standard treatment for diffuse intrinsic brainstem tumours as surgery is not an option in the majority of patients.The target is the macroscopic lesion on MRI (T1 weighted with contrast medium) with a safety margin. Although the majority of patients (~70 per cent) show a significant improvement in neurologic status following such treatment, the prognosis is very poor. The median time to disease progression is of the order of 5–6 months, the median survival time is <1 year, and survival at 2 years and beyond is <10 per cent. These findings have led to dose-escalation studies with hyperfractionation (55–78 Gy, 1.2 Gy twice daily) by the Paediatric Oncology Group (POG), the Children's Cancer Group (CCG), the University College of San Francisco (UCSF), and the Children's Hospital of Philadelphia (CHOP). No clear survival advantage could be proved in this relatively large cohort of children. High doses (>72 Gy) led to more radiation-related brain damage (e.g. intralesional necrosis). Results of further (randomized) trials (e.g. POG) have to be awaited before recommendation of any aggressive hyperfractionated high-dose radiotherapy for standard treatment.29 In the POG trial survival for patients treated with hyperfractionated radiotherapy plus cisplatin was worse than for patients receiving the same radiotherapy alone.30

The principles of chemotherapy are identical to the protocols described above for glioblastoma.


Medulloblastoma, the most frequent malignant paediatric brain tumour, is a distinctive embryonal brain tumour originating in the posterior fossa, locally infiltrating in the brainstem or the fourth ventricle, or growing continuously along the cerebrospinal pathway. Tumour cell dissemination into the CSF is detected at diagnosis in a quarter of the patients.

The diagnosis of medulloblastoma is usually suspected from preoperative MRI. Dissemination via the CSF must be investigated by MRI and CSF cytology before starting postoperative therapy. A surgically based staging system attempts to classify tumour stage (T1–T4) (location, volume, and extension into neighbourhood structures), and stage of metastatic disease (M0–M4). Disease classified as T1–T3a M0 is regarded as early stage (favourable), and T3b–4 is regarded as locally advanced high stage (unfavourable).

Some prognostic factors are commonly described in series dealing with medulloblastoma. The most significant adverse factor is CSF spread at diagnosis (M1) and presence of macroscopic metastatic disease (M2–M4). Further important prognostic factors are the age of the child, with worse outcome in young children (< 2 years), and tumour resectability, which is correlated with local stage.

Surgery and postoperative radiotherapy are the standard treatments, but nowadays chemotherapy also plays an important role for all patients. In patients with low and average risk disease the use of chemotherapy has allowed a reduction in the dose of radiotherapy to the craniospinal axis and appears to have brought about a significant improvement in disease-free and overall survival. Patients with high-risk disease now fare better with combined chemoand radiotherapy.4

Surgery aims at total tumour removal, which can now be achieved in the vast majority of cases (including ‘near-total’ removal). However, in general, no major surgery-related permanent neurologic deficits are acceptable.

After surgery a ‘posterior fossa syndrome’ (e.g. truncal ataxia, cerebellar mutism) may occur. This may be transient but may also take months to remit. Its presence should not delay postoperative treatment.31

Postoperative CSI with 35 Gy (less in low risk patients with chemotherapy) and an additional field to the posterior fossa with a total dose >50 Gy using 1.6–1.8 Gy per fraction is currently the gold standard for cure of medulloblastoma. New studies with hyperfractionated radiotherapy have attempted to increase the dose to the tumour region up to a high total dose without increasing the risk for late effects. A dose of 1 Gy twice daily to the whole cerebrospinal target, CSI up to 24–36 Gy, and a local dose up to 68–72Gy are being evaluated in different studies.32,33

There is some controversy about the adequate dose for prophylactic irradiation (spinal and supratentorial). In series with a final outcome of >50 per cent 5-year survival, the most frequently reported doses have been about 35 Gy. In favourable cases, radiation dose reduction alone (without additional chemotherapy) has led to inferior results with early isolated neuraxis relapse and lower 5-year event-free and overall survival than standard irradiation (36 Gy).34 In disseminated disease the standard dose of 36 Gy is suboptimal; therefore the CSI dose should be increased to 40 Gy, with additional boost doses of up to 45–50 Gy.35 Careful design of target volumes is mandatory (CSI and posterior fossa boost), based on all the information available (e.g. surgical and histopathologic reports, MRI) and on institutional experience. CSI is one of the most complicated radiation treatment techniques in a young child. Poor results may be due to inadequate radiation treatment planning and performance, and therefore great care should be taken with this treatment.

Medulloblastoma is a chemosensitive tumour. Chemotherapy (platinum, vincristine, CCNU, alkylating agents) is becoming increasingly accepted as standard in all patients with medulloblastoma during first-line treatment. Five-year progression-free survival rates are better after combined chemoand radiotherapy than after radiotherapy alone. Many trials are currently under way, trying to clarify chemotherapy schedules and the timing of their delivery within a combined treatment approach (CCG, POG, SIOP, GPOH, and CHOP). The current HIT 2000 protocol for medulloblastoma without metastasis is shown in Figure 17.3. Bone marrow or stem cell transplantation with high-dose chemotherapy is a treatment option for high-risk patients.

Fig. 17.3 Design of medulloblastoma HIT 2000 trial for tumours without metastasis.

In various series, survival rates ranging from 50 to 70 per cent at 5 years, and from 30 to 60 per cent at 10 years, have been reported after surgery and radiotherapy. Single institutions even report survival rates of 80 per cent at 5 years.1,36


PNETs arising in the supratentorial region are associated with a different outcome, even when treatment similar to that given for medulloblastoma is used. These tumours often occur between birth and 5 years. They are found predominantly in the cerebral hemispheres, most commonly in the frontal and temporal lobes. On imaging, they may appear as wellcircumscribed masses, but there is often widespread microscopic extension. Glial, neuronal, or ependymal differentiation may be seen. Leptomeningeal spread is uncommon compared with medulloblastoma and is less frequent at presentation than with medulloblastoma. Patients present with non-specific signs of raised intracranial pressure, seizures, or motor impairment. Investigations should include imaging of the whole ventricular system and spine to exclude seeding. Surgical resection is often difficult because of their size and position. Craniospinal irradiation has been recommended, but results in progression-free survival rates of only 30 per cent. Chemotherapy follows the guidelines for medulloblastoma.


Ependymomas arise from ependymal cells lining the ventricles and the central canal of the spinal cord. Nearly two-thirds occur in the infratentorial compartment. Anaplastic tumours tend to be more common in the posterior fossa than in the supratentorial region of young children. Tumour cells in the CSF are described in 5–15 per cent of patients; the rate of drop metastases is higher in anaplastic ependymoma. Prognostic factors are the age of the patient, tumour location, tumour stage, histology, extent of resection, and radiation dose.

MRI of the brain and spinal cord at the time of diagnosis are mandatory for optimal treatment planning. Surgery, radiotherapy, and chemotherapy are used in treatment. Surgery is the most effective treatment, but is rarely curative. There are three indications for surgery: for tumour resection at diagnosis, for resection of residual disease after adjuvant therapy, and for resection of a relapse, usually at the primary site. The primary goal is to remove the tumour totally and re-establish CSF flow. If significant morbidity is to be avoided, total removal is often not possible in posterior fossa tumours because of the specific growth pattern of ependymomas with invasion of the fourth ventricular floor and encasement of lower cranial nerves and regional arteries.

Although the role of postoperative radiotherapy is well established, there are many controversies regarding the appropriate extent of radiotherapy. Currently, local postoperative radiotherapy (55 Gy, 1.8–2 Gy per fraction) to the primary tumour site with broad margins (>2 cm) is considered appropriate not only for low-grade and all supratentorial ependymoma but also for anaplastic infratentorial ependymomas. CSI is the standard basic treatment only for disseminated tumours.19 At present, local dose escalation is being prospectively evaluated if there is postoperative residual disease.

The role of chemotherapy is unclear. In current clinical trials, chemotherapy with vincristine, cisplatin, etoposide, and cyclophosphamide is being investigated. There seems to be an indication for children aged <3 years, but neither the St Jude study nor ‘Baby POG’ has shown improved long-term outcome.

Overall, 5-year survival rates, mostly after combination treatment, range from 28 to 60 per cent depending on prognostic factors.1,36 The frequent local recurrence (within 18–24 months of diagnosis) is almost always incurable. Nevertheless, local palliative treatment procedures may improve symptoms for the limited life expectancy (Fig. 12.4).

Choroid plexus neoplasms

Choroid plexus neoplasms occur in all age groups but represent 10–20 per cent of brain tumours seen in children during the first year of life. They more commonly arise in the lateral ventricles in children and are capable of secreting CSF. The dominant presenting feature are signs and symptoms of raised intracranial pressure. Papillomas and carcinomas are capable of leptomeningeal dissemination.

The treatment of benign choroid plexus papillomas is total surgical excision, which is usually curative. Because of the haemorrhagic tendency of the tumour, operative mortality still remains high. A more aggressive and anaplastic tumour, the choroid plexus carcinoma, accounts for 10–20 per cent of choroid plexus neoplasms. The likelihood of achieving gross total resection is less in this tumour because of local invasion, and therefore the likelihood of cure is limited.

A pilot study evaluating the feasibility of an intercontinental phase III chemotherapy study for patients with choroid plexus tumours is active (CPT-2000). According to this protocol, patients with choroid plexus papilloma will be followed without adjuvant treatment and reoperated on if there is recurrence. Patients in whom further surgical resection is impossible and patients with choroid plexus carcinoma are randomized between a carboplatin-based and a cyclophosphamide-based protocol. Both arms will also include vincristine and VP16. Since the majority of patients are infants, they will not receive irradiation. However, irradiation will be given in both treatment arms after the second block of chemotherapy to patients >3 years of age.

Fig. 17.4 Recurrence of ependymoma: distribution of radiation dose from stereotactic radiotherapy using multiple arcs.

Pineal tumours and germ cell tumours

Tumours arising in the region of the pineal gland are rare, and can be classified into three groups: germ cell tumours, pineal parenchymal tumours, and gliomas.

Primary CNS germ cell tumours (40–65 per cent of all pineal tumours) mainly arise in the pineal or suprasellar region. Their incidence is substantially higher in Japan than in Western countries. Their peak age incidence is in the second or third decades of life. Germ cell tumours are pathologically heterogeneous, reflecting varied cell types of origin. At least half to twothirds are pure germinomas. The remaining patients have non-germinomatous tumours, which may be classified, depending upon their predominant constituent cell type(s), into embryonal carcinoma, endodermal sinus (or yolk-sac) tumour, choriocarcinoma, malignant teratoma, or malignant mixed germ cell tumours. The histologic appearance of these tumours is identical to similar tumours occurring outside the CNS. Rare patients are found with immature or mature teratomas without any malignant germ cell elements.

Once the presence of a pineal region mass has been demonstrated radiographically, CSF and serum should be examined for tumour markers such asα-fetoprotein (AFP) and the β subunit of human chorionic gonadotrophin (β-HCG), which are produced by intracranial yolk-sac tumours and choriocarcinoma, respectively. This is important, because in patients with significant marker elevation (β-HCG > 50 IU 1-1; AFP > 25 ng ml-1), surgery is not necessary to establish a tissue diagnosis before starting therapy. In all other patients, however, a histologically verified diagnosis is mandatory. Benign teratomas are often surgically completely resectable, and thereby cured. Leptomeningeal seeding may be present in patients with malignant germ cell tumours, and spinal MRI and CSF cytology are required for clinical staging.

For pure germinomas, radiotherapy alone leads to cure rates of up to 100 per cent (e.g. GPOH trial32). However, dose (40–50 Gy local treatment and 22–35 Gy CSI) and volume remain issues of debate. Some groups (e.g. SFOP) use platinum-based multidrug regimens in combination with local radiotherapy and achieve slightly worse results.

Non-germinomatous germ cell tumours are rarely cured with radiotherapy alone, and the addition of platinum-based chemotherapy is essential. Preoperative chemotherapy has been shown to be effective in facilitating complete resection of large or infiltrating tumours and in diminishing the risk of a primary operation. However, local radiotherapy is important for local control, even in patients with complete response to chemotherapy.

Pineal parenchymal tumours consist of pineoblastomas and pineocytomas. Pineoblastomas develop in the younger patient and are PNETs. The high malignancy suggests that the approach to treatment should be similar to medulloblastoma and include aggressive resection, CSI with a local boost, and chemotherapy. Pure pineocytomas are slower-growing, relatively well-circumscribed masses, and have a benign course. Complete surgical removal represents the definitive treatment. Careful evaluation of the histology must exclude malignant components revealing a mixed tumour, which should be treated according to its most malignant part.1


Craniopharyngiomas are histologically benign tumours originating in the sellar and suprasellar area from embryonic squamous cell rests of the pharyngeal–hypophyseal duct. Their macroscopic and neuroradiologic appearance is of a cystic, solid, or mixed-mass lesion, typically with calcifications. Since they are slow-growing tumours, they may reach large sizes before they become symptomatic, with endocrine dysfunction, visual problems, or signs of ICP with obstructive hydrocephalus in 30 per cent.

The best management of craniopharyngioma in children remains controversial. Treatment options for this tumour include microsurgery, fractionated conformal radiotherapy, stereotactic procedures, intracavitary radiotherapy, and intracavitary chemotherapy.

Surgery is performed with either an attempt at radical total removal or more conservatively, if necessary followed by radiotherapy. Total removal is possible in 60–90 per cent of cases. However, even in these recent series, recurrence rates between 7 and 34 per cent are reported, with an average of 23 per cent, and major morbidity remains a concern even in experienced hands. Morbidity relates not only to a deterioration of visual and pituitary function, but even more importantly to complex and severe cognitive and neurobehavioural disturbances. If the tumour is totally removed, no further immediate treatment is indicated.37

Recurrence is treated by reoperation followed by radiotherapy, or by radiotherapy alone. As 70 per cent of patients with partial resection will show tumour progression, these patients should be irradiated using conformal or stereotactic techniques (55–60 Gy). Occasionally, radiation will be deferred in a young child, but these patients require close follow-up. With a definitely lower mortality and morbidity, the long-term tumour control with this combined approach compares well with initial aggressive total resection alone, with 20 per cent recurrent tumour growth.

Intracavitary radiotherapy with 32Por 90Y in experienced centers has given excellent results in monocystic tumours. Intracystic application of bleomycin has not only been shown to facilitate subsequent resection due to fibrosis of the capsule, but can also lead to shrinkage and long-term control.

Intramedullary spinal cord tumours

Spinal cord tumours account for only 3–6 per cent of primary CNS tumours in children. Of intramedullary tumours, up to 60 per cent are astrocytomas and 20–30 per cent are ependymomas. Oligodendrogliomas and gangliogliomas are less frequent. In general, these tumours are well-differentiated low-grade tumours, with only 10 per cent having high-grade or anaplastic features. The tumours may be focal or extend to involve multiple segments. The most common symptom is local pain along the spinal axis, alteration of a previously normal gait, and other signs of spinal cord malfunction occurring later in the course of the disease. Hydrocephalus may complicate the clinical picture in as many as 15 per cent of patients with spinal cord tumours.

Surgery for diagnosis and resection, if possible, is mandatory. Complete surgical resections are difficult in astrocytomas and appear to be possible more frequently in ependymomas which tend to have a clearer cleavage plane. Postoperative orthopaedic follow-up and monitoring for spinal deformity are important.

No controlled trial of radiation or chemotherapy has been carried out for intramedullary tumours, and evidence for their usefulness has to be inferred from the treatment of similar tumours in other CNS sites, such as cranial low-grade glioma and ependymomas. If radiotherapy is performed, it is targeted to the tumour region, with a total dose of 40–50 Gy, and seems to improve functional recovery.

The overall survival rates of low-grade astrocytomas with various degrees of resection and postoperative radiation therapy are 66–70 per cent at 5 years and 55 per cent at 10 years. Local recurrence rates as high as 33–86 per cent have been reported. In ependymomas, survival rates, depending on the amount of resection, vary from 50 to 100 per cent at 5 years and 50 to 70 per cent at 10 years.38

In the rare anaplastic or high-grade tumours, postoperative total neuraxis irradiation and adjuvant chemotherapy are recommended. However, patients usually succumb to their disease because of local progression or dissemination along CSF pathways.

Management of brain tumours in very young children

Brain tumours in infants and very young children have unique properties with regard to clinical presentation, anatomical location, histology, and prognosis that distinguish them from brain tumours occurring in the older child. By the time of diagnosis, most tumours in infants are quite large.

Delay in diagnosis occurs, in part, because the infant skull can expand to accommodate raised ICP, hence masking for some time the typical signs and symptoms associated with a mass lesion. Infants may present with failure to thrive (despite good appetite and food intake), endocrinopathies, developmental delay, vomiting, decreased visual acuity, and nystagmus accompanied by expanding head circumference.

Historically, infants with brain tumours have had the worst prognosis of any age group. Although delay in diagnosis, tumour type and size, and the tendency for early dissemination may be important factors, the poor outcome probably reflects the limitations of treatment.

Surgery is more difficult in the young child because of tumour size, the fragility of the immature brain, and problems related to neuroanaesthesia. In addition, radiation is known to be very toxic in this age group. Therefore, the dose is routinely reduced by at least 10–20 per cent. This reduced dose is probably inadequate for tumour control. Nevertheless, even with reduced radiation dose, major long-term effects, including learning disabilities, mental retardation, endocrinopathies, leucoencephalopathy, and vasculopathy, are to be expected in a significant number of patients. These concerns have resulted in interest in delaying or eliminating radiation in this young population by using postoperative chemotherapy.

The largest study of prolonged postoperative chemotherapy and delayed radiation in infants with brain tumours has come from the POG.39 A total of 198 children aged <3 years with malignant brain tumours were treated with a combination of cyclophosphamide, vincristine, cisplatin, and VP16. Brainstem gliomas and pineoblastomas showed little or no response. The 5-year progression-free survival was 30 per cent and the overall 5-year survival was 39.4 per cent, which compared favourably with historical controls using standard postoperative radiation. Except for ependymomas, which tended to progress after several years, most failures occurred during the first 6 months of chemotherapy. While studies over the past decade have clearly confirmed that medulloblastomas are chemosensitive and that even leptomeningeal disease can be effectively treated at times, the POG data on ependymomas suggest that these tumours are chemosensitive but not chemocurable. Therefore, unlike young children with good-risk medulloblastoma, i.e. those who have had a gross tumour resection and no metastases, from whom radiation can be withheld, radiation may be deferred but usually cannot be eliminated in children with ependymoma. However, it is hoped that radiation-induced neurotoxicity to the hypothalamus–pituitary region, temporal lobes, and auditory apparatus can be reduced by the use of focal conformal radiation techniques, thereby potentially limiting the damage to growth, hearing, and cognition.

Another approach to prevent the need for craniospinal irradiation that deserves further evaluation is the use of high-dose chemotherapy with either autologous bone marrow transplant or stem cell rescue.

Late effects in children treated for brain tumours

As 5and 10-year survival rates of children with CNS tumours have increased, so has concern over late effects of treatment. Many long-term survivors have intellectual, endocrine, and neurologic deficits that lead to significant social handicaps as well as diminished quality of life. Damage to the CNS from several sources may play a role in these deficits. Direct destruction of normal brain tissue by tumour, as well as surgical trauma, may cause some degree of irreversible neurologic damage. Likewise, chemotherapy, especially in combination with radiation, appears capable of inducing encephalopathy. However, it is radiation therapy that has been implicated as the main cause of adverse long-term sequelae, particularly intellectual impairment. Some reports suggest that most children receiving whole-brain radiation have some form of cognitive deficit in various intelligence quotients, visual/perceptual skills, learning abilities, and adaptive behaviour. Prospective controlled studies have found a younger age at diagnosis, radiotherapy, methotrexate chemotherapy, tumour location, and time interval to testing to be important and related to a high risk of subsequent cognitive deficits.40,41 Dose, fractionation, and volume of radiation influence the development of these deficits, with more severe sequelae occurring at higher doses and larger volumes.19 Only recently, vasculopathies and multiple cerebral cavernomas have been described as relevant late effects of radiotherapy.42,43 Thus, current cooperative group studies are treating infants and very young children with brain tumours with prolonged postoperative chemotherapy in an attempt either to delay or to eliminate cranial radiation entirely.

Detailed studies have revealed a wide range of endocrine dysfunction following cranial irradiation which includes the hypothalamic–pituitary region. The most common impairment is growth failure due to growth hormone deficiency.1 Although growth hormone replacement therapy will improve longitudinal growth, it has not been as effective as in children with idiopathic growth hormone deficiency. Another factor contributing to decreased growth is spinal irradiation. The younger the child at the time of spinal irradiation, the more severe are the adverse effects on vertebral body growth.

Hypothyroidism may also occur, and if not corrected may lead to poor linear growth, learning difficulties, and lethargy. Evaluation of thyroxine and thyroid-stimulating hormone function will allow early treatment of this problem. Gonadal dysfunction has only recently been reported in children with brain tumours. Radiation to the sacral spine and a number of cytotoxic drugs, particularly alkylating agents, are associated with gonadal damage and can cause ovarian failure, oligospermia, or azoospermia in these patients. As more chemotherapy is used and the patients are followed longer, it is likely that a much higher incidence of these side effects will be noted. Thus several risk factors need to be addressed by future studies, and careful planning of drug–radiation combinations is essential to maximize survival while reducing long-term local and systemic sequelae. Therefore it is of paramount importance that these children remain under long-term surveillance so that problems are anticipated and therapeutic strategies instituted as early as possible (see also Chapter 7). Secondary tumours have been observed, but exact data are lacking.


1. Cohen ME, Duffner PK (eds) (1994). Brain Tumours in Children. Principles of Diagnosis and Treatment (2nd edn). New York: Raven Press, 127–46, 177–201, 219–39, 329–46, 455–81.

2. Thomas DGT, Graham DI (eds) (1995). Malignant Brain Tumours. New York: Springer-Verlag.

3. Sposto R, Ertel IJ, Jenkin RD, et al. (1989). The effectiveness of chemotherapy for treatment of high grade astrocytoma in children: results of a randomized trial. A report from the Children's Cancer Study Group. J Neurooncol 7, 165–77.

4. Freeman CR, Taylor RE, Kortmann RD, Carrie C (2002). Radiotherapy for medulloblastoma in children: a perspective on current international clinical research efforts. Med Pediatr Oncol 39, 99–108.

5. Kleihues P, Cavenee WK (eds) (2000). Pathology and Genetics of Tumours of the Nervous System. Lyon: IARC. 6. Pollack IF (1994). Brain tumours in children. N Engl J Med 331, 1500–7.

7. Segal RA, Goumnerova LC, Kwon YK, et al. (1994). Expression of the neurotrophin receptor trkC is linked to a favorable outcome in medulloblastoma. Proc Natl Acad Sci USA 91, 12867–71.

8. Grotzer MA, Janss AJ, Fung K-M, et al. (2000). TrkC expression predicts good clinical outcome in primitive neuroectodermal brain tumors. J Clin Oncol 18, 1027–35.

9. Herms J, Neidt I, Lüscher B, et al. (2000). cmyc expression in medulloblastoma and its prognostic value. Int J Cancer 89, 395–402.

10. Grotzer MA, Hogarty MD, Janss AJ, et al. (2001). MYC messenger RNA expression predicts survival outcome in childhood primitive neuroectodermal tumor/medulloblastoma. Clin Cancer Res 7, 2425–33.

11. Gilbertson RJ, Perry RH, Kelly PJ, et al. (1997). Prognostic significance of HER2 and HER4 coexpression in childhood medulloblastoma. Cancer Res 57, 3272–80.

12. Pomeroy SL, Tamayo P, Gaasenbeck M, et al. (2002). Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 415, 436–42.

13. Kraus JA, Felsberg J, Tonn JC, et al. (2002). Molecular genetic analysis of the TP53, PTEN, CDKN2A, EGFR, CDK4 and MDM2 tumour-associated genes in supratentorial primitive neuroectodermal tumours and glioblastoma of childhood. Neuropathol Appl Neurobiol 28, 325–33.

14. Raffel C (1996). Molecular biology of paediatric gliomas. J Neurooncol 28, 121–8.

15. Gurney JG, Severson RK, Davis S, Robison LL (1995). Incidence of cancer in children in the United States. Cancer 75, 2187–95.

16. Koos WT, Miller MH (1971). Intracranial Tumors of Infants and Children. Stuttgart: Thieme, 14–22.

17. Barkovoch AJ (1993). Paediatric neuroimaging. Contemporary Neuroimaging (2nd edn), Vol 1. New York: Raven Press.

18. Berger MS (1996). The impact of technical adjuncts in the surgical management of cerebral hemispheric low-grade gliomas of childhood. J Neurooncol 28, 129–55.

19. Halperin EC, Constine LE, Tarbell NJ, Kun LE (1994). Paediatric Radiation Oncology (2nd edn). New York: Raven Press, 40–139.

20. Kun L (1994). Principles of radiotherapy. In: Cohen ME, Duffner PK (eds) Brain Tumours in Children. Principles of Diagnosis and Treatment, (2nd edn). New York: Raven Press, 95–116.

21. Kaatsch P, Rickert CH, Kühl J, et al. (2001). Population-based epidemiologic data on brain tumors in German children. Cancer 92, 3155–64.

22. Michaelis J, Kaletsch U, Kaatsch P (2000). Epidemiology of childhood brain tumors. Zentralbl Neurochir 61, 80–7.

23. Campbell JW, Pollack IF (1996). Cerebellar astrocytoma in children. J Neurooncol 28, 223–32.

24. Janss AJ, Grundy R, Cnaan A, et al. (1995). Optic pathway and hypothalamic/chiasmatic gliomas in children younger than age 5 years with a 6-year follow-up. Cancer 75, 1051–9.

25. Packer RJ, Ater J, Allen J, et al. (1997). Carboplatin and vincristine chemotherapy for children with newly diagnosed progressive low grade gliomas. J Neurosurg 86, 747–54.

26. Prados MD, Edwards MSB, Rabbitt J, et al. (1997). Treatment of pediatric low grade gliomas with a nitrosourea-based multiagent chemotherapy regimen. J Neurooncol 32, 235–41.

27. Walker DA, Punt JAG, Sokal M (1999). Clinical management of brain stem glioma. Arch Dis Child 80, 558–64.

28. Lyden DC, Mason WP, Finlay JL (1996). The expanding role of chemotherapy for supratentorial malignant gliomas. J Neurooncol 28, 185–91.

29. Heideman RL, Packer J, Albright LA, et al. (1997). Tumours of the central nervous system. In: Pizzo PA, Poplack DG (eds) Principles and Practice of Paediatric Oncology (3rd edn) Philadelphia, PA: JB Lippincott, 633–98.

30. Freeman CR, Kepner J, Kun LE, et al. (2000). A detrimental effect of combined chemotherapyradiotherapy approach in children with diffuse intrinsic brain stem gliomas. Int J Radiat Oncol Biol Phys 47, 561–4.

31. Sutton LN, Phillips PC, Molloy PT (1996). Surgical management of medullo-blastoma. J Neurooncol 29, 9–21.

32. Kortmann RD, Kuhl J, Timmermann B, et al. (2001). Current and future strategies in interdisciplinary treatment of medulloblastomas, supratentorial PNET (primitive neuroectodermal tumors) and intracranial germ cell tumors in childhood. Strahlenther Onkol 177, 447–61.

33. Prados MD, Wara W, Edwards MS, et al. (1996). Treatment of high-risk medulloblastoma and other primitive neuroectodermal tumors with reduced dose craniospinal radiation therapy and multi-agent nitrosourea-based chemotherapy. Pediatr Neurosurg 25, 174–81.

34. Thomas PR, Deutsch M, Kepner JL, et al. (2002). Low-stage medulloblastoma: final analysis of trial comparing standard-dose with reduced-dose neuroaxis irradiation. J Clin Oncol 18, 3004–11.

35. Jenkin D (1996). The radiation treatment of medulloblastoma. J Neurooncol 29, 45–54.

36. Kaye AH, Laws ER Jr (eds) (1995). Brain Tumours. Edinburgh: Churchill Livingstone, 493–504, 561–74, 665–671.

37. Epstein FJ, Handler MH (eds) (1991). Craniopharyngioma: the answer. Paediatr Neurosurg 21 (Suppl 1), 1–130.

38. Epstein FJ, Constantini S (1995). Spinal cord tumours of childhood. In Pang D (ed) Disorders of the Pediatric Spine. New York: Raven Press, 371–88.

39. Duffner PK, Horowitz ME, Krischer JP, et al. (1999). The treatment of malignant brain tumors in infants and very young children: an update of the Pediatric Oncology Group experience. Neurooncology1, 152–61.

40. Radcliffe J, Bunin GR, Sutton LN, et al. (1994). Cognitive deficits in long-term survivors of childhood medulloblastoma and other non-cortical tumors: age-dependent effects of whole brain irradiation. Int J Dev Neurosci 12, 327–34.

41. Mulhern RK, Reddick WE, Palmer SL, et al. (1999). Neurocognitive deficits in medulloblastoma survivors and white matter loss. Ann Neurol 46, 834–41.

42. Siffert J, Allen JC (2000). Late effects of therapy of thalamic and hypothalamic tumors in childhhood: vascular, neurobehavioural and neoplastic. Pediatr Neurosurg 33, 105–11.

43. Heckl S, Aschoff A, Kunze S (2002). Radiation-induced cavernous hemangiomas of the brain: a late effect predominantly in children. Cancer 94, 3285–91.

If you find an error or have any questions, please email us at Thank you!