27 Tumours of the brain and skull

Over the last 10 years, there there have been several important advances in cell biology, molecular genetics, and targeted therapies in neuro-oncology. Improved neurosurgical techniques such as frameless stereotaxy, awake craniotomy, and intra-operative MRI, safer methods of directing radiotherapy, new chemotherapy approaches, and novel modalities of therapy provide optimism that there will eventually be some improvements in treatment-related morbidity and survival. There has also been an increasing change from individual clinician decision making to decision making by multidisciplinary teams of neurosurgeons, neurologists, clinical oncologists, neuropathologists, neuroradiologists, and specialist nurses with the aim of improving decision making, management planning across specialties, communication, and enrolment in suitable clinical trials. In addition, Good Clinical Practice guidelines, an international ethical and scientific quality standard for designing, conducting, recording, and reporting trials, increases the onus and responsibilities on clinical investigators to perform trials to the highest standard and to have the trials externally monitored, and the trial conduct and results audited. While these obligatory and statutory responsibilities are labour intensive and time consuming, they should improve the quality of trials by limiting the possibility of unintentional bias or fraud. Improving the recording of serious adverse event reporting through trial quality assurance and quality control procedures will help ensure that a balanced view of the effects of a drug or procedure is identified earlier than in the past. It will be interesting to see how research develops over the next decade.

Tumours of the brain and skull can be divided by site and relationship to the cranial fossae:

Histological grading is crucial to the understanding of brain tumours and their eventual management. The World Health Organisation classification (Table. 27.1) (Kleihues et al. 1993a) recognizes that tumours develop via different pathways and that the behaviour of some grades of malignancy are quite distinct. The classification also allows standardization of reporting for epidemiological studies. The World Health Organization have revised the grading system of central nervous system tumours to take into account recently defined histological entities such as pleomorphic xanthoastrocytoma and dysembryoplastic neuroepithelial tumours (Table. 27.2). Identification of the type of tumour and predictions of the prognosis and response to treatment can be refined by genotyping methods of identifying genetic changes common in certain tumour types, for instance Ch 1p and 19q loss or methyl guanine methyl transferase, MGMT, promoter methylation status.

Brain tumours are the second most common cause of death from neurological disease, surpassed only by stroke. Neuroepithelial brain tumours are the fifth most common cause of death from malignancy under the age of 65 years. If other intrinsic brain tumours, including metastases and primary central nervous system lymphomas are considered, it is very likely that intracranial tumours are one of the three most common causes of death from cancer in the working population.

Information from cancer registry statistics show that neuroepithelial tumours are the most common solid malignancy in children, the fourth most common in the under-45 age group, and the eighth most common under the age of 65 years. However, cancer registries frequently do not reliably record incidence data on all primary intracranial tumours, particularly meningiomas, acoustic neuromas, pituitary adenomas, and primary central nervous system lymphoma (Counsell et al. 1997).

Meningiomas, acoustic neuromas, and pituitary tumours account for almost 25 per cent of all intracranial tumours (Table. 27.3), and there is evidence that more tumours are being identified, although this may reflect the wider availability of brain imaging. Primary central nervous system lymphomas associated with AIDS, have also increased. Brain metastases affect 20 per cent of patients with cancer and account for over 50 per cent of all

intrinsic brain tumours. The incidence of intracranial metastases may also be increasing, as the brain will also act as a ‘sanctuary site’ for metastases as treatment of the systemic components of the tumour improves.

The incidence of all intracranial tumours is between 21 to 30 cases per 100 000 population each year (Counsell et al. 1996; Counsell and Grant 1998; Pobereskin and Chadduck 2000) (Table. 27.3). Assuming no demographic or regional differences, one would expect approximately 200 000 new cases of intracranial tumour each year in Europe and 100 000 new cases each year in North America. Over the last 50 years the incidence of intracranial tumours in developed countries has increased (McKinney 2004). The increase is predominantly in the elderly population and may be due to better case ascertainment and improved health care for the elderly and the shifting demographics to a more elderly population (Radhakrishnan et al. 1995). Studies produce a wide range of incidences for neuroepithelial tumours, 2.5 to 9.1/10⁵/year, but studies with good case ascertainment are at the higher figure (Kaye et al. 1993; D’Alessandro et al. 1995). The age specific incidence of primary brain tumours has a small peak in early childhood, 3.5/10⁵ cases/year, dips in the 15 to 24 years age range, at 2.9/10⁵/year, and then progressively climbs to its highest incidence in the 65 to 74 years age range of 24/10⁵/year before falling off slightly to 9.6/10⁵/year in the >85 years age range (Counsell et al. 1996). Cerebral metastases are rare in the under 25 years age group, at 1/10⁵/year, but increase gradually to 53.7/10⁵/year in the 65 to 74 years (CI 41.6–68.2/10⁵/year.) age group. Meningeal tumours are also very rare under 25 years of age, at 0.4/10⁵/year and increase steadily to a peak incidence of 9.0/10⁵/year in the age range 75 to 85 (CI 3.6–18.5/ 10⁵/year). A meta-analysis of incidence studies, suggests that the differences could be accounted for by methods of identifying patients, inclusion criteria for what represented primary intracranial tumours, and the scanning techniques used to identify cases. Methodological guidelines for future incidence studies in brain tumours have been suggested, in order to standardize future incidence studies to allow meaningful analysis (Counsell and Grant 1998).

Medulloblastomas and ependymomas occur most frequently in childhood or early adolescence. Grade 2 astrocytomas and oligodendroglioma in early adults aged 20–40 and grade 3 and 4 astrocytomas, the malignant gliomas, most commonly in the >40 age range although they can occur at any age.

Patients with intracranial tumours can present to a variety of hospital-based clinicians or primary-care physicians as a result of increased availability of direct access CT scanning (Grant et al. 1996). Following imaging diagnosis, further specialist referral is almost always the rule. Referral occurs at a time of great uncertainty for the patient and family, and close collaboration between clinicians is essential to reduce anxiety. A clear explanation to the patient and family and a simple well constructed management plan must be given, which should be consistent between specialists. Recently, multidisciplinary team meetings have being advocated to discuss individual patients and advise on the best management plan. These meetings are often tumour site specific, so for pituitary tumours they involve endocrinologist, neurosurgeons, clinical oncologist, ophthalmologist, and specialist nurse (NICE: http://www.nice.org.uk/page.aspx?o=282278). Communication with patient, family, and general practitioner should be unambiguous and supportive (Davies and Hopkins 1997a).

Males are 1.1 to 1.7 times more likely than females to develop primary intracerebral brain tumours (Counsell et al. 1996) and females are about 2.2 times more likely to develop meningiomas (Counsell and Grant 1998). There do not appear to be any specific geographical factors that influence the incidence of primary brain tumours (McKinney 2004, Wrensch et al. 2002). Some studies report that white Americans have a higher incidence of brain tumours than blacks, Hispanics, and Japanese Americans; however, others have only shown a higher incidence in black children than whites (Bunin 1987) and have not found a higher incidence in whites compared with Hispanics.

The level of affluence is associated with the likelihood of developing neuroepithelial brain tumours in adults and children. The incidence of neuroepithelial tumours is approximately 2.3 times greater in areas of SE Scotland with the greatest affluence than with the least affluent (McLoone 1993). The relationship between deprivation category and incidence of neuroepithelial tumours is linear (Counsell et al. 1996). The incidence of metastases was about twice as high in the least affluent groups than the most affluent (X₂ trend 6.18, p=0.01). The relationship between neuroepithelial tumours and social deprivation is very unlikely to be due to inequality of access to care in the health service because of the National Health Service and because the inverse relationship is seen with cerebral metastases.

Inherited cancer syndromes and very rare cases due to therapeutic radiation are the only causative factors that have been unequivocally identified (IARC 2003). The strongest possible risk factor is therapeutic cranial irradiation. Therapeutic irradiation for scalp ringworm was associated with a 4-fold increase in the risk of brain tumour (Ron et al. 1988). A Japanese study of atomic bomb survivors failed to demonstrate any increased risk of brain tumours in children exposed to radiation in utero (Kato et al. 1989). Certain chemicals such as polycyclic aromatic hydrocarbons and nitroso compounds can induce brain tumours in experimental animals, but their role in human brain tumours remain unproven (Maekawa and Mitsumori 1990). Vinyl chloride can cause brain tumours in rats and a Swedish study showed that workers in a vinyl chloride plant were twice as likely to die from brain tumours (Hagmar et al. 1990). There is no good evidence that cellular telephones are associated with brain tumours. Dietary and other chemical exposures, maternal birth characteristics, head trauma, infections, vaccinations and various medications, including anti-nauseants during pregnancy, have been suggested as risk factors, but none have shown strong enough correlations to confirm a link (Wrensch et al. 2002). Immunodeficiency states are a strong risk factor for developing primary central nervous system lymphoma.

Many different genes have been associated with familial cancer predisposition and have associations with central nervous system tumours (Table. 27.4). There is a genetic predisposition to brain tumours in patients with the family cancer syndrome (Li and Fraumeni 1969). Li–Fraumeni syndrome has been linked to a mutation in p53 on chromosome 17p13 (Malkin et al. 1990).

p53 mutations have been frequently found in all grades of astrocytoma and represent an early event in malignant transformation of astrocytic cells (Ohgaki et al. 1993). Glioblastomas with p53 mutations generally have arisen from a low grade astrocytoma and are often called ‘secondary’ glioblastomas. Glioblastomas that have arisen de novo, i.e. not preceded by a low grade glioma, generally do not have a p53 mutation and are called ‘primary’ glioblastomas.

In hereditary syndromes with linked gene mutations such as tuberous sclerosis at 9q32–34, about 5 per cent of cases will develop cerebral gliomas. Tuberous sclerosis is an autosomal dominant condition and genes that are involved are TSC1 gene on Ch 9q34 and TSC2 gene on Ch 16p13. The incidence is thought to be between 1 and 2/10,000.

Neurofibromatosis type 1, another autosomal dominant condition with near complete penetrance, is related to mutations at Ch 17q11–22 and is associated with cerebral or optic nerve glioma in between 4 and 45 per cent of patients. In neurofibromatosis type 2, with mutations at Ch 22q12, schwannomas, ependymomas, and meningiomas are found more frequently than in the general population. Although neurofibromatosis type 1 is an autosomal dominant disorder with near complete penetrance, about half of neurofibromatosis type 1 cases are new mutations. Neurofibramatosis type 2 is also autosomal dominant.

Von Hippel–Lindau disease is an autosomal dominant condition characterized by the development of haemangioblastomas of the central nervous system and other sites. It is due to a defect in the tumour suppressor gene on 3p25–26. Up to 50 per cent of patients develop cerebellar haemangioblastomas and 15 per cent develop spinal haemangioblastomas.

Other rare syndromes such as Gorlin’s syndrome, Turcot’s syndrome, and Gardner’s syndrome are associated with brain tumours. Gorlin’s syndrome is an autosomal dominant condition with naevoid basal cell carcinoma and is associated with medulloblastoma. Turcot’s syndrome and familial polyposis is associated with an increased incidence of glioma and Gardner’s syndrome with medulloblastoma.

Our knowledge of cellular and molecular biology has expanded exponentially over the last 10 years (Rasheed et al. 1999). The developments have been used to confirm the histological nature of tumours, determine the proliferating potential of malignant cells and provide insight into the control mechanisms behind tumour cell proliferation and tumour suppression. Three classes of genes have been implicated in the pathogenesis of cancer: oncogenes, tumour suppressor genes, and mismatch repair genes. Oncogenes are derived from normal cellular genes, or proto-oncogenes, which function as growth stimulators. However, if activated, uncontrolled cellular proliferation occurs. Tumour suppressor genes normally restrain cellular proliferation but when inactivated by gene mutation, will result in uncontrolled cellular proliferation. Knudsen proposed the ‘Two hit hypothesis’ which suggested that the loss of both alleles of the tumour suppressor gene was necessary before a cancer could develop. Patients with a germline mutation in one of these alleles require a somatic change in the other allele before a tumour develops. The third class of gene important to the development of tumours are the mismatch repair genes. The products of these genes identify mismatches that occur at point mutations in the genome and repair these. If there is a problem with failure to identify mismatches, gene repair and control are affected. Carcinogenesis may be initiated by a mutation or alteration in certain genes, for instance deletion or rearrangement of DNA, or by several different changes in several genes. The mutation can be random, inherited, or induced by radiation, chemicals, or other insult.

Growth factors are usually under the control of proto-oncogenes and tumour suppressor genes. Several proto-oncogenes have been associated with the development of malignant gliomas, such as c-erb B-1, c-sis, c-myc, ras, c-fos, and ros oncogenes. Erb-b encodes for epidermal growth factor receptor, EGFR, and c-sis encodes for platelet derived growth factor, PDGF. The most common growth factor in malignant glioma is EGFR. EGFR is over-expressed in about 40 to 60 per cent of malignant gliomas but expression is uncommon in astrocytoma. This suggests that the EGFR occurs as a late feature of deregulation of cell division and not an initiating factor of tumourogenesis. Amplification or over-expression of EGFR is associated with resistance to chemotherapy-induced apoptosis. There is, however, not a close relationship between over-expression of EGFR and survival in tumours of the same grade. PDGF receptor amplifications are common in low-grade astrocytomas, suggesting that actroption of PDGF may be an early event in pathogenesis of malignant astrocytoma. Several other growth factors such as transforming growth factors, TGF-alpha and TGF-beta, fibroblast growth factor, and insulin-like growth factor are also over-expressed in some glioblastomas. TGF-alpha is a polypeptide growth factor that has sequence homology for EGF and binds to EGFR. TGF-alpha is frequently amplified in anaplastic astrocytoma and glioblastoma multiforme and may act as a growth factor ligand for EGFR and form an autocrine growth loop leading to proliferation of glioma cells. The amplification and over-expression of TGF-alpha might be an earlier event in a gradual process of tumourigenesis. TGF-beta stimulates astrocytoma cells to migrate and invade (Yamada et al. 1995).

The Bcl-2 protein family may contribute to impaired ability of glioma cells to undergo apoptosis. Bcl-2 controls apoptosis through release of cytochrome C from mitochondria and activation of caspases, which initiate the cell death process.

Several different chromosomes have been reported as having possible loci for tumour suppressor genes. Low grade brain tumours frequently demonstrate loss of genetic material on chromosomes 6, 13, 17p, or 22. P53 gene is the most well known tumour suppressor genes and is situated on chromosome 17p13.1. Mutations of the p53 gene are implicated in many cancers. In gliomas, dysfunction of key components of the apoptotic pathways, prevent apoptosis of damaged cells and contribute to tumour development and malignant progression. In response to DNA damage, wild-type p53 induces G1-S cell cycle arrest and either mediates DNA repair or initiates apoptosis if the damage is too great to repair. Inactivation of p53 is an early event in glioma development and it promotes genomic instability by allowing highly damaged cells to survive and accumulate further mutations. As the grade of malignancy increases, other allelic changes are seen in addition. Allelic loss of Ch 19q is found in 46 per cent of grade 3 gliomas, but only 11 per cent of grade 2 tumours (von Deimling et al. 1994). The most common chromosomal abnormality, present only in glioblastoma, is loss of chromosome 10. A region on chromosome 10q23.3 called PTEN acts as a tumour suppressor gene. PTEN alterations occur with a high frequency in primary glioblastoma, usually in elderly patients where the tumour arises de novo, but also occur with a low frequency in secondary glioblastoma, usually in young patients where the glioblastoma develops from a low grade to a high grade tumour. In primitive neuro-ectodermal tumours, PNET-medulloblastomas and PNET variants, deletions of the short arm of chromosome 17p, or duplication of the long arm 17q occur in a high percentage of cases and there is frequently also loss of information on chromosome 10, 11, and 22. Seventy per cent of meningiomas have loss on chromosome 22. Despite the large number of potential tumour suppressor genes, there are no specific losses of genetic material that are unique to a particular histological group of tumours and no specific deletion or other genetic alteration is found in all tumours.

Patients with tumours of the head and neck most commonly present with headache, lower cranial neuropathies, or facial pain. Where the involvement is in the nasopharynx, one should consider nasopharyngeal carcinoma, adenoid cystic carcinoma and metastatic tumours.

Where there is involvement of the carotid body region, glomus jugulare should be considered. Patients with glomus jugulare tumours commonly present with lower cranial nerve palsies, such as dysphonia, dysphagia with wasted tongue, or weak palate (Section 20.7). If extensive, a glomus jugulare can also cause pulsatile tinnitus, headaches, and hearing loss. They are usually large when eventually discovered. Plain X-rays, with a submento-vertex view, will best demonstrate the enlargement of the jugular foramen. CT scanning shows the strongly enhancing tumour mass with erosion of the adjacent bone. MRI will demonstrate the ‘salt and pepper’ appearance caused by the flow voids within the tumour. Gadolinium enhanced coronal MRI scan is particularly useful to delineate the extent of the tumour and the relationship to the brain stem.

Where there is occipital headache, and lower cranial neuropathies, tumour involvement of the skull base should be considered. In the midline, chordoma, chondroma, and chondrosarcoma are all possible. Imaging of the head and neck by CT scan or MRI may demonstrate a lesion with or without soft tissue involvement. CT scanning is superior when bony involvement is present, as in skull osteomas, Paget’s disease and fibrous dysplasia, and MRI with its multiplanar capabilities is superior for soft tissue visualization, as in nasopharyngeal carcinoma, metastases, and glomus jugulare tumours. Contrast enhancement will better delineate blood vessels from surrounding soft tissue structures and otorhinolaryngological assessment, angiography, and simple blood tests, such as alkaline phosphatase and myeloma screen, may also be helpful.

Patients with olfactory groove meningiomas usually present late when the tumour is large enough to cause headache, seizures, or personality changes. Anosmia is a rare complaint in the absence of other symptoms.

Pituitary tumours may present because of endocrine effects, hormone excess or hypopituitarism, or mass effect. Women with prolactinomas commonly present with amenorrhoea and galactorrhoea and are referred to gynaecologists. Men with prolactinomas usually present later than women and may complain of headache, reduced sexual function and visual field defects. The presence of acromegaly or steroid excess will point to a diagnosis of a growth hormone or ACTH secreting macroadenoma respectively and should stimulate a request for imaging of the pituitary region. Plain lateral skull X-rays may show a ‘double floor’ or erosion of the sella tursica in the presence of a pituitary macroadenoma. Skull X-ray, however, may be entirely normal in macroadenoma and is always normal in microadenomas and is insufficient to exclude any intracranial tumour. Investigations for hormone secreting pituitary tumour include prolactin levels, growth hormone levels, and serum or urinary cortisols plus imaging of the pituitary gland by multiplanar contrast enhanced MRI, or contrast enhanced coronal CT with fine cuts through the pituitary gland (Fig. 27.1).

Craniopharyngiomas commonly present with symptoms and signs of mild hypopituitarism or diabetes insipidus. Almost 90 per cent of men complain of impotence, while most women complain of amenorrhoea. Children present with short stature and 40 per cent of patients will be hypothyroid at presentation while 25 per cent have adrenal insufficiency. Fifty per cent of patients will have diabetes insipidus and headache. Craniopharyngiomas are usually a complex combination of cysts and solid tumour with calcification (Fig. 27.2). There is generally no surrounding oedema in the brain. Craniopharyngiomas can be difficult to differentiate from dermoid or epidermoid tumours or Rathke’s pouch cysts, but craniopharyngiomas generally have more complex cysts than epidermoids, thicker irregular walls, and more calcification on CT scan. Hypopituitarism is also found in 20 per cent of patients with epidermoid and dermoid cysts in the suprasellar or parasellar areas (Fonari et al. 1990).

Patients with visual field loss due to optic nerve or chiasm compression will be referred to ophthalmologists, physicians, or neurologists. The visual impairment may be due to pathology in the nerve, for instance glioma, or pressure on the nerve from a tumour, such as pituitary tumour, craniopharyngioma, meningioma, or metastasis, or a cyst such as dermoid, epidermoid, or Rathke’s pouch. Symptoms will lead to imaging of the anterior visual pathway. Multiplanar gadolinium enhanced MRI is the investigation of choice.

The coronal images will provide useful information about expansion of the optic nerves consistent with an optic nerve glioma or meningioma, and about the parasellar region and the relationship with any extrinsic pressure on the optic chiasm. Pituitary macroadenomas usually cause expansion of the pituitary fossa and displace the optic chiasm upwards, producing a bitemporal field loss which starts in the superior temporal quadrants. The visual field defects with pituitary macroadenomas will vary depending on whether the optic chiasm is prefixed or postfixed. Craniopharyngiomas expand downwards from the hypothalamus and cause a bitemporal hemianopia most frequently involving the inferior temporal quadrants, but can also cause a variety of visual field defects depending on where the tumour presses on the visual apparatus. Epidermoid and dermoid cyst generally appear on CT scanning as well-circumscribed lesions with low density between that of CSF and brain due to cholesterol or keratin granules. The wall of epidermoid cysts may be thinly calcified, and since the contents are avascular they do not enhance with contrast. Dermoids are more heterogeneous, have a thicker wall, rarely enhance, and more commonly demonstrate calcification. There is generally no surrounding oedema in the brain. On T1-weighted MRI sequences, epidermoids exhibit a variable signal, white when the lipid content is high or black if lipid content is low. Classically, epidermoids have low signal on T1-weighted images and very high signal on T2-weighted images. Dermoids give high signal on T1-weighted images in areas containing fat. They give variable signal where there is a combination of fat, muscle, and bone and may be mistaken for a craniopharyngioma or mixed germ cell tumour, or teratoma. Mixed germ cell tumours are more heterogeneous than germ cell tumours, or germinomas, because they contain a variety of tissues including bone, cartilage, hair, and fatty tissue. Enhancement following contrast is common in germ cell tumours. Tumours such as hypothalamic astrocytomas and oligodendrogliomas can also extend downwards to cause chiasmal or optic nerve compression and hypopituitarism. These tumours are usually solid with areas of calcification but can also sometimes be exophytic. Rathke’s cleft cysts are simple intrasellar cysts containing CSF. Meningiomas are usually easily differentiated from cysts and other tumours, however, en plaque meningioma of the optic nerve may be difficult to visualize even with gadolinium enhanced MRI and should always be considered as a potential diagnosis in patients with progressive optic nerve disease even in the absence of a clear mass lesion on MRI.

Patients who present with peri-orbital pain, ocular muscle paralysis, or ptosis may have tumours in the orbit, often metastases or lymphoma, lacrimal gland carcinoma, or tumours of the sphenoid wing, usually meningioma, carcinoma, dermoid, epidermoid, large pituitary tumours, or craniopharyngioma. Differential diagnosis will depend on the speed of onset of symptoms and the imaging appearance.

If the presenting complaint is facial numbness or weakness, deafness, tinnitus, or vertigo, patients are likely to be sent by their family practitioners to see a physician or otorhinolaryngologist. The differential diagnosis includes acoustic neuroma, meningioma, haemangioblastoma, dermoid, epidermoid, lipoma, and metastasis.

The most common cerebellopontine angle tumour is an acoustic neuroma (Fig. 27.3). Patients most commonly present with deafness, tinnitus, or vertigo. Patients with meningiomas less commonly have acoustic nerve symptoms and more commonly present with other cranial nerve involvement, especially facial numbness and facial weakness, however, differentiation on clinical grounds is unreliable. MRI is most sensitive imaging technique to delineate lesions of the middle or posterior cranial fossae. Acoustic neuromas usually expand the acoustic nerve and may cause expansion of the internal auditory meatus. Small tumours enhance uniformly and are usually easy to distinguish from other tumours; however if acoustic neuromas are very large it may be difficult to identify the origin of the tumour and distinguish it from a meningioma. Meningiomas strongly enhance uniformly on CT reflecting the vascularity of these tumours, but necrosis, cysts, and calcification can alter the signal characteristics on MRI making differentiation from dermoids or even haemangioblastomas rather difficult. Cholesteatomas and epidermoids can commonly be differentiated from meningiomas and acoustic neuromas by their relative lack of enhancement.

If the lower cranial nerves are involved it is imperative to get good imaging of the base of the skull, neural exit foramena and extracranial soft tissues in the neck.

Headache, memory or personality changes, and seizures are the most common initial symptoms in patients with primary intracerebral tumours. However, patients are commonly referred to hospital only when focal symptoms or signs become obvious, such as seizures, hemiparesis papilloedema, dysphasia, or hemianopia. Hemiparesis or hemisensory loss are the most common symptoms at hospital presentation (Table. 27.5). Nearly all patients who have weakness or numbness complain of these symptoms, thus directing the clinician to the abnormality on examination. Only 7 per cent of patients with malignant glioma complain of visual symptoms, yet over 20 per cent have signs of visual field loss and 23 per cent have papilloedema; therefore, careful examination of the visual fields and fundi is important in anyone complaining of headaches or symptoms suggestive of disturbance of higher mental function. Most commonly, the upper motor neurone weakness is mild initially and affects fine manipulation first and mild progressive lower limb weakness, involving hip flexion, knee flexion, and ankle dorsiflexion (Fig. 27.4). Clinical follow up using quick sensitive simple tests such as the timed nine hole peg test, timed 10 m walk, and a test of memory and grading of dysphasia are usually

sufficient to assess clinical response to treatment (Grant et al. 1994; Clyde et al. 1998). The Barthel Activities of Daily Living Index (Section 6.5.2) may be useful in elderly patients or in patients with metastases where the weakness is severe, but is insensitive and the ‘ceiling effect’ precludes its use in trials of glioma in general and does not record cognitive disability or dysphasia. The Karnofsky Performance scale is useful for grading patients for entry into studies, but in practice is usually used a 3 point grading scale (70, 60–50, <50) rather than an 11 point scale (100, 90, 80,…, 10, 0). It can be used to follow individual patients although intra-observer and inter-observer errors limit its usefulness.

Stroke-like onset or collapse with coma, occurs in about 5 per cent of patients with intracerebral tumours and is most commonly related to haemorrhage into a malignant glioma or metastasis. Stroke-like presentations and subacute presentations with cognitive deficits, visual field disorders, or dysphasia are more common in the elderly and most commonly suggest a poor prognosis.

Late onset epilepsy, defined as first seizure after age 18, is a common presentation in patients who have low grade gliomas and meningiomas. It has been estimated that between 3 per cent and 10 per cent of patients with late onset epilepsy have an underlying tumour of some form. Seizures are the first presenting symptom in 54 per cent of low grade gliomas, 50 per cent of anaplastic astrocytomas, 26 per cent of meningiomas, 19 per cent of glioblastomas, 15 per cent of metastases and 11 per cent of primary central nervous system lymphomas. Over a follow-up period of 3 years, the prevalence of seizures rises to 70 per cent in low grade glioma, 56 per cent in anaplastic astrocytoma, 48 per cent in glioblastoma, 44 per cent in meningioma, 39 per cent in primary central nervous system lymphoma, and 31 per cent in metastases. Tumour associated epilepsy is focal in approximately 50 per cent of patients, partial epilepsy with secondary generalization in 25 per cent, and tonic–clonic seizures without warning in 25 per cent of patients.

Children are more likely to have posterior fossa or deep thalamic region tumours and present with cerebellar symptoms; symptoms of raised intracranial pressure are more frequent.

Early identification of patients with intra-cerebral tumours is important but very difficult. Various attempts to draw up referral guidelines for patients with intracerebral tumours have been devised (http://www.nice.org.uk/page.aspx?o=cg027niceguideline), which include a timescale by which such patients should be seen and scanned.

CT and MR brain scanning has dramatically improved the management of patients with brain tumours, but diagnostic interpretation is not without its difficulties. The three levels of diagnosis are: is it a tumour; if so, what type of tumour is it; and if it is a glioma, what grade of glioma is it?

Histopathological correlation has proved that enhancing areas on post-contrast CT and MRI scans correspond to densely cellular, hypervascular tissue of viable tumour (Fig. 27.5) (Burger et al. 1983; Whelan et al. 1988). A consistent finding from stereotactic biopsy studies is of a variable zone of microscopic tumour infiltration outside the enhancing area, extending at least as far as the abnormal signal on the T2-weighted images. There may be even greater extension of isolated tumour cells beyond these radiologically defined boundaries (Burger et al. 1988; Kelly et al. 1987).

The small uniform cells of a primary central nervous system lymphoma are densely packed together. These tumours are commonly more radio-dense than surrounding brain on non-contrast enhanced CT scan, giving the appearance that some contrast has been given (Fig. 27.6). When contrast is given they can enhance intensely. Although characteristic, these features are not specific for primary central nervous system lymphoma and can be seen in other tumours comprising of tightly packed cells, as in some metastases, some small cell malignant gliomas, and medullobastomas. Frequently in AIDS patients, primary central nervous system lymphomas can demonstrate ring enhancement similar to an abscess. Primary central nervous system lymphomas are commonly located in a periventricular distribution deep in the white matter and approximately two-thirds occur in the cerebral hemispheres and one-third infratentorially.

The pineal gland lies between the splenium of the corpus callosum above and the superior colliculus below. Numerous tumour types can arise in the pineal region: germ cell tumours, pineocytomas, gliomas, metastases, cysts. Patients commonly present with Parinaud’s syndrome with limitation of upgaze, convergence, and impaired pupillary reaction to light and accommodation. Eyelid retraction, Collier’s sign, or ptosis can also occur. Some patients will have diabetes insipidus and if there is a beta-human chorionic gonadotrophin, beta HCG, secreting germinoma precocious puberty can occur. CT or MRI scan with gadolinium will demonstrate the lesion which is generally causing an obstructive hydrocephalus. The most common tumours in this area are teratomas or germinomas. Pineocytomas are less common and arise from the pineal parenchymal cells. Gliomas can arise in the pineal region and metastases to the pineal region also occur. In addition to these solid tumours, pineal region cysts can occur and can be simple, filled with CSF, or can be epidermoid or dermoid cysts. Germinomas are usually solid, enhance uniformly, and may be surrounded by calcium. Teratomas also enhance but are heterogenous with multiloculated cysts. Non-germinomatous germ cell tumours are heterogenous and may enhance irregularly or not at all. Choriocarcinoma bleeds frequently and intratumoural haemorrhage or bleeding into a cyst may occur. Pineocytomas have low signal on T1-weighted MRI, high signal on T2-weighted images and enhance uniformly, but there can be calcification within the tumour, or haemorrhage if there are any pineoblastoma elements. Pineal gliomas can vary on imaging depending on whether they are low or high grade tumours. Tumours at this site quite commonly seed to the spinal canal, therefore full spinal MR imaging is also important. Serum and CSF, obtained in the absence of hydrocephalus, analysis for germ cell markers can be valuable. Malignant teratoma, germinoma with syncytioblastic cells, embryonal carcinoma and, endodermal sinus tumours may have elevated levels of alphafetoprotein. Choriocarcinoma and embryonal carcinoma may have elevated levels of beta HCG, whereas only germinomas and germinomas with syncytiotrophoblastic cells have elevated levels of placental alkaline phosphatase. Non germ cell tumours are negative for all these markers. CSF histology is usually negative. In the presence of these hormones, the most likely diagnosis is a teratoma or choriocarcinoma. If the CSF is negative for these markers, then biopsy of the pineal tumour is recommended. Shunting or fenestration of the ventricles may be necessary prior to definitive operation or at the time of operation. Transventricular biopsies run the risk of bleeding from the veins surrounding the anterior pineal region. Resection of the tumour from behind requires retraction of the occipital lobe and is frequently associated with temporary or permanent visual field defects.

It is wise not to be overconfident about the suspected type or grade of a tumour when discussing the situation with the patient and relative prior to biopsy or resection, for all the reasons mentioned above regarding diagnostic certainty. The main classification scheme used to type and grade tumours is the World Health Organization classification (Table. 27.1). Management strategies and prognosis are considerably changed depending on the tumour type and tumour grade. For example the diagnostic workup and management of certain types of tumours, such as primary central nervous system lymphoma, medulloblastoma, germinoma are completely different from that for others such as glioma, meningioma, or metastasis. Some tumour types are very sensitive to radiation, for instance germinoma, lymphoma, whilst others, such as glioma, teratoma, are strongly resistant to radiation. Some tumours are chemosensitive, for instance lymphoma, germinoma, and anaplastic oligodendroglioma, while others are relatively chemoresistant, such as glioblastoma or meningioma.

The main histological characteristics used to grade primary malignancy are: increased cellularity, increased number, and atypical appearance of mitotic activity, pleomorphism or anaplasia, proliferation of the vascular endothelium, and tissue necrosis. Grading systems such as Dumas–Duport, have been validated by correlation between the level of anaplasia and survival after diagnosis (Kim et al. 1991).

Previously, a key diagnostic criterion for glioblastoma multiforme was the presence or absence of necrosis, but the revised World Health Organization system, WHO-2, and the Dumas–Duport systems do not require necrosis for classification as glioblastoma multiforme (Kleihues et al. 1993b; Dumas–Duport et al. 1988). Necrosis did predict a shorter survival, however, the magnitude of the survival difference between patients with and without necrosis was small, with median survival 10.9 months versus 12.5 months respectively. This supports the WHO-2 classification of endothelial proliferation without necrosis as sufficient to make a diagnosis of glioblastoma multiforme (Barker et al. 1996). Most centres use the WHO system or occasionally the Dumas–Duport system.

The heterogeneous nature of some tumours, means that insufficient sampling of the tumour can lead to undergrading of tumours and result in inappropriate treatment planning. The use of image directed stereotactic techniques has led to a 95 per cent diagnostic success rate, even in small tumours in sites with difficult access, and allows several samples from representative areas of the tumour to be taken so that grading is more reliable (Bernstein and Parrent 1994). Interobserver concordance is also variable in these heterogeneous tumours: glioblastoma multiforme–89 per cent; anaplastic astrocytoma–77 per cent; oligoastrocytoma–48 per cent; oligodendroglioma–39 per cent (Aldape et al. 2000). Genotyping of these tumours will better define their nature.

Tumours of the head and neck can be divided based on their position in relation to the skull and, if biposied, by their cell type. Extra-cranial tumours of the head and neck are usually due to carcinoma of the paranasal sinuses, nasopharynx, oral cavity, or oropharynx or are a result of tumours arising from local structures, such as blood vessels in the case of glomus jugulare, or bone in chordoma or chondroma. These tumours commonly produce cranial neuropathies.

Head and neck cancer usually affects patients in the fifth and sixth decades. The most common primary tumours are carcinoma of:

Squamous cell carcinoma is the most common cell type. Risk factors are smoking and possibly alcohol. Up to 10 per cent of patients with head and neck tumours have second malignancies which also tend to be cancers associated with tobacco or alcohol, originating in lung, oesophagus, or stomach.

Tumours of the nasal cavity and sinuses have an incidence of <1/10⁵/year, affect men twice as often as women and usually occur in patients over 60 years. Higher incidences are found in Japan and South Africa. Occupational factors are important such as nickel or chromium exposure, radium and isopropyl alcohol. The frequency of nasal and paranasal cancers is higher in furniture, shoe, and textile industries (Roush 1979). Tumours in this region are commonly squamous cell carcinomas, but esthesioneuroblastomas and adenoid cystic tumours can also be aggressive and invade the cranial cavity. Tumours are frequently far advanced by the time of diagnosis, because the symptoms of nasal blockage or discharge are frequently ignored. Extension to the orbit may cause diplopia and proptosis and extension upwards may result in direct invasion of the cribriform plate and frontal lobes with anosmia and headache. Surgery usually involves a combined approach from a skull base team comprising neurosurgeon, otorhinolaryngologist, and faciomaxillary surgeon. Factors that make surgical resection difficult or sometimes impossible include the involvement of the base of skull, nasopharynx, or sphenoid sinus. In inoperable patients, local disease can be controlled in a substantial proportion of patients using radiation therapy.

Nasopharyngeal carcinoma is endemic in China and North Africa. Dietary factors such as nitrosomines in salt-cured food and viral factors, particularly Epstein–Barr virus, have been implicated in the malignant transformation into nasopharyngeal cancer. These tumours usually occur predominantly in 40 to 60 year olds and males are twice as likely to be affected. Nasopharyngeal carcinoma most commonly present as a lump in the neck, but can also cause nasal blockage and deafness from blockage of the eustachian tube or invade the skull base or cavernous sinus causing cranial nerve palsies. Surgical biopsy confirms the diagnosis and radiation therapy is the treatment of choice. At recurrence, surgery has a limited role. Chemotherapy can be helpful to treat widespread metastatic disease, which occurs in a high proportion of patients with this tumour. Cisplatinum, methotrexate, and epirubicin all have some palliative effect. Five year survival rates of 50 to 60 per cent can be achieved with combinations of these treatment modalities.

Tumours of the oral cavity are typically squamous and a close link with smoking has been identified. Alcohol may also be implicated in the aetiology. Diagnosis is usually earlier than other head and neck tumours, and neural involvement is less common. These tumours can however cause the ‘numb-chin syndrome’ due to involvement of the mental nerve, a branch of the mandibular division of the trigeminal nerve (Section 20.1.6).

Tumours of the oropharynx usually occur in patients older than 50 years and are 4-fold commoner in males. Smoking and alcohol are the most common risk factors. Presentation is commonly quite late and lymph node spread and neural involvement are frequently found. Bulbar palsy or isolated involvement of the lower cranial nerves are not uncommon. Surgery and radiation therapy are the mainstays of treatment and survival depends on the site, extent, and staging of the tumour. Laryngeal and hypopharyngeal carcinomas cause local pain or referred pain to the ear and lymphadenopathy in the neck. Dysarthria and swallowing difficulties and recurrent laryngeal nerve palsy can also occur.

In general, if patients with head and neck tumours have early disease, radical resection and post-operative radiotherapy can be curative, although the cosmetic and functional result of treatment may affect quality of life. In advanced disease, where cure is not possible, it may be best to treat the primary site by radiation and any loco-regional lymph nodes by surgical resection and radiotherapy. The 5 year survival for nasal and sinus carcinoma is approximately 60 per cent, but depends on the extent of primary disease at presentation. Chemotherapy in the initial treatment regime can lead to significant tumour regression in 60 to 90 per cent of cases. A systematic review of concomitant radiotherapy and chemotherapy in patients with locally advanced head and neck cancer revealed an overall survival benefit (OR=0.62; 95 per cent CI: 0.52–0.74; p <0.00001; RR=0.83, RD=11 per cent) although combined treatment produced more adverse events than radiation therapy alone (Browman et al. 2001). Meta-analyses of individual patient data from 31 randomized controlled trials of neoadjuvant chemotherapy for patients with locally invasive head and neck cancer and meta-analyses of adjuvant chemotherapy revealed no significant survival advantage over loco-regional treatment alone (Pignon et al. 2000).

Chordomas are very rare slow growing tumours that arise from remnants of the notocord and generally occur in the midline in the region of the clivus at the base of the skull. They are locally invasive and usually present late because of the non-descript nature of the chronic headache or neck pain. They may present with intermittent diplopia, facial numbness if the upper clivus is involved, or as a nasopharyngeal mass with multiple lower cranial neuropathy, affecting glossopharyngeal, vagus, accessory, and hypoglossal nerves if the lower clivus is involved. Most clivus chordomas produce destruction of the clivus and have extradural extension. CT scan best demonstrates bone destruction but MRI better demonstrates the tumour margins and soft tissue structures and blood vessels. MRI demonstrates low signal intensity on T1-weighted scans. Chordomas rarely calcify or enhance. MRI demonstrates low signal intensity on T1-weighted scans. Imaging alone cannot adequately differentiate between chordoma, chondroid chordoma, and chondrosarcoma; however, chondrosarcomas usually arise along the petro-occipital fissure and more commonly calcify. Occasionally chondrosarcomas can arise in the midline in the region of the clivus.

Because of the approximation of the tumour to sensitive neural and vascular tissues, complete surgical removal is not usually possible but surgery is required to make the diagnosis and to reduce bulk disease. Care should be taken to keep the dura intact when performing trans-sinus surgery because of the risk of post-operative meningitis; however, since 50 per cent of these tumours have already breached the dura by the time of diagnosis, careful post-operative packing with fat and muscle grafts is essential. Radical resection of anterior skull base tumours has improved (Lawton et al. 1995) in the hands of teams of surgeons specializing in skull base surgery, but the advice is to follow surgical removal with local radiation therapy. There are no randomized studies of the effect of radiation and conventional photon irradiation has shown no dose–response relationship (Tai et al. 1995). Despite this there have been advances in radiosurgical methods, using gamma knife, linear accelerator, and proton beam that have been purported to give better local control and spare surrounding tissues (Schulz-Ertner et al. 2002; Krishnan et al. 2005). Chemotherapy, even with multi-agent chemotherapy is only rarely effective (Scimeca et al. 1996). The natural history of chordoma is minimally affected by surgery and radiotherapy, although symptom control can be achieved adequately. Five year survival rates range from 10 to 58 per cent in different series, but most series suggest that <50 per cent of patients are alive at 5 years. Systemic metastases occur in between 10 and 30 per cent.

Glomus tumours arise from the glomus jugulare and can spread medially to involve the middle ear or skull base and therefore can present with hoarseness and dysphagia due to lower cranial nerve palsies (Section 20.7.3). In some cases these tumours can cause pulsatile tinnitus and deafness and the tumour can be seen behind the tympanic membrane. CT or MRI scan is usually sufficient to make the diagnosis, but angiography is also valuable to define the blood supply to the tumour and to embolise the tumour pre-operatively. Surgery can be hazardous and usually involves a skull base team. Surgery requires mastoid and a sub-occipital craniectomy and can be complicated by meningitis or facial nerve palsy. The place of radiation therapy is debatable. The tumour is benign, but complete resection is often not possible. Local control can be achieved by radiation therapy alone in approximately 90 per cent of patients with inoperable glomus tumours (Springate and Weichselbaum 1990).

Fibrous dysplasia is a benign fibrous process that can involve the skull vault or the base of the skull. The condition can present in isolation or as part of the McCune–Albright syndrome of fibrous dysplasia, cafe au lait spots, and endocrinopathy comprising precocious puberty, thyrotoxicosis, primary hyperparathyroidism, or hyerprolactinaemia. McCune–Albright syndrome is due to post-zygotic somatic mutations in the gene encoding G alpha s proteins, GNAS1 (Ringel et al. 1996). Pathologically, the bone is replaced by fibrous tissue composed of collagen and fibroblasts. It can be difficult to distinguish fibrous dysplasia from an ossifying fibroma. Headaches or cranial nerve involvement are the common presenting symptoms. Craniobasal fibrous dysplasia can produce progressive visual loss due to extradural optic nerve compression if it affects the bones of the orbit or conductive and sensorineural hearing loss if the disease affects the temporal bone. A canal cholesteatoma is found in 40 per cent of patients with temporal bone involvement by fibrous dysplasia. Surgical resection is usually the treatment of choice. Radiation therapy may also be beneficial in relieving symptoms. Malignant transformation to a fibrosarcoma is rare but can occur.

Cranial osteomas are common and not usually symptomatic. Occasionally because of their size or site they can cause cosmetic disfigurement, headache, cranial neuropathies, or seizures. Rarely they are associated with Gardner’s syndrome. A new classification for cranial osteomas has been suggested (Haddad et al. 1997). The classification divides osteomas into parenchymal, dural, skull base, and skull vault, with the latter being divided into enostotic and exostotic variants. CT scan is the imaging technique of choice. Indications for surgery include progressive ophthalmoplegia, neurological signs, and significant cosmetic deformity.

Cranial Paget’s disease, osteitis deformans, can be monoostotic or polyostotic. Paget’s disease involves the cranial vault and temporal bones in 30 to 40 per cent of cases. It is characterized by excessive and disorganized bone formation and resorption. Patients may present with headaches, hearing loss, tinnitus and vertigo, or hemifacial spasm. The bone alkaline phosphatase is almost always elevated and is a useful marker, but the best marker to follow cranial involvement with Paget’s is the serum propeptide carboxyterminal of type 1 procollagen for new bone formation and the urinary C-terminal telopeptide of type 1 collagen for bone resorption (Alvarez et al. 1997). Calcitonin and etidronate and newer bisphosphonates, such as pamidronate, alendronate can help control the turnover of new bone and help control pain and prevent neurological or orthopaedic complications (Selby et al. 2002).

Intracranial extracerebral tumours arise from pituitary gland, a pituitary adenoma, meninges, a meningioma, nerves, a neuroma, or neuroepithelial remnants, a dermoid or epidermoid, and are usually benign. Metastases directly to the dural space do occur, especially with breast cancer and lymphoma, but frequently they are a late complication in patients with widespread systemic involvement or as part of a malignant meningitic process where they can produce mass lesions.

Pituitary tumours can arise from any cell in the anterior pituitary gland, account for 10 per cent of intracranial tumours and have an incidence of 2 to 3/100 000/year. They are most commonly benign adenomas, although pituitary carcinomas and metastases to the pituitary do occur. Tumours can produce symptoms and signs through hypersecretion of hormones or related to the mass effect of the tumour on the optic nerves or structures adjacent to the pituitary gland. Tumours generally have to be >1cm before they cause any symptoms of neural or vascular compression. When the tumour expands the pituitary gland it can cause hypofunction of certain hormones and produce hypothyroidism, amenorrhoea, or Addison’s disease. Involvement of the neurohypophysis can result rarely in diabetes insipidus. When the tumour extends laterally it can involve the cavernous sinus and result in cranial neuropathies. Haemorrhage into the pituitary gland can cause pituitary apoplexy.

The most common type of pituitary adenoma is the microadenoma. Microadenomas by definition are less than 10 mm in diameter, do not expand the pituitary fossa and usually secrete prolactin. About 70 per cent of prolactin secreting tumours are microadenomas.

Pituitary metastases are infrequent and usually occur in the context of known systemic cancer with metastatic spread to other sites, but occasionally can be the only manifestation of metastatic spread. Most cases present with headache or diabetes insipidus. Panhypopituitarism and visual field impairment is demonstrated in about 25 per cent (Sioutos et al. 1996).

Pituitary tumours can be radiologically classified based on their size and growth characteristics. Grade 0 is where there are no imaging abnormalities, grade I is where there are minor changes in the pituitary but the tumour is <1 cm, grade II is where there is diffuse enlargement but no focal sellar destruction, grade III is where there is local invasion of the sella and grade IV is where there is extensive destruction of the sella. Further subclassification will depend on the extent and direction of supra or parasellar extension. Coronal and sagittal MRI with gadolinium contrast gives high definition information about the pituitary, parapituitary region, and adjacent soft tissues; however it is not as good as CT at demonstrating bone erosion. High resolution contrast enhanced coronal CT scan with 1.5 mm contiguous slices will also provide information about the homogeneity of the pituitary gland and its relationship to surrounding structures. CT of a microadenoma will characteristically show low density within the gland; however, small cysts in the pituitary occur in normal people and therefore the diagnosis of microprolactinoma should not be made on radiological grounds alone. Characterization of the tumour by MRI plus or minus MR angiography is usually sufficient to plan surgery (Fig. 27.1). Formal arterial angiography is not usually required.

Management Recommendations for service provision and guidelines for management have been developed (Royal College of Physicians 1997).

Dopamine agonists There is general agreement that bromocriptine is the treatment of choice in prolactin secreting adenomas. Bromocriptine is a dopamine agonist that directly stimulates specific pituitary cell membrane dopamine D2 receptors and inhibits prolactin synthesis and secretion. Treatment will usually cause a reduction in size of any macroadenoma and reduce or normalize prolactin levels in blood. There have been reports of macroadenomas enlarging despite treatment with bromocriptine and close review of visual acuity and fields is recommended. If patients are unable to tolerate bromocriptine because of side effects, other dopamine agonists can be tried, including pergolide, cabergoline, or lisuride.

Pregnancy During pregnancy there can be enlargement of the normal pituitary gland by up to 70 per cent and therefore special care has to be taken during pregnancy in patients with pituitary tumours, especially macroadenomas. During pregnancy bromocriptine can be discontinued if the tumour is a microadenoma. Ideally, it is advisable to wait till any macroadenoma has been adequately treated before a planned pregnancy is attempted. Bromocriptine should be restarted after pregnancy or during pregnancy if there is neurological deterioration.

Surgery Prolactin secreting macroadenomas can also be treated by surgical resection, if there is serious visual compromise or side effects with medical treatment. Visual improvement occurs in 80 per cent of cases but prolactin levels frequently do not reduce to normal. The recurrence rate of macroadenomas after surgery ranges from 25 per cent to 75 per cent, therefore radiotherapy after surgical resection may be necessary if drug therapy is not possible. Neurosurgery or radiotherapy is sometimes recommended before pregnancy particularly for macroadenomas that have not responded to bromocriptine.

Non-functioning adenomas are usually macroadenomas. Drugs are ineffective at reducing the tumour size. Surgery is the treatment of choice and most adenomas can be dealt with via a transphenoidal approach. Surgery will confirm the diagnosis and relieve compression by the tumour on surrounding structures. About 75 per cent of patients with visual field defects will have some recovery of visual fields post-operatively and headache is relieved in over 90 per cent (Ebersold et al. 1986; Comtois et al. 1991). Post-operative complications occur in <10 per cent of cases and include diabetes inspidus, CSF leak, transient hyponatraemia, meningitis, and sinusitis. Diabetes insipidus in the immediate post-operative period occurs in 10 to 20 per cent of patients but is permanent in only 2 to 5 per cent. It is uncommon to achieve complete resection in macroadenomas because most have invaded the dura or surrounding structures. Post-operative imaging can be performed at 2 to 3 months and close follow up by MRI will usually identify tumour regrowth and determine which patients may benefit from radiotherapy. Some centres advocate radiation therapy to any residual tumour, but the frequency of radiation induced complications should be considered especially in younger patients.

Pituitary Cushing’s syndrome can be difficult to diagnose. High dose dexamethasone suppression test, using 2 mg dexamethasone 6 hourly for 48 hours, with suppression of the morning serum cortisol to <50 per cent basal value, points to pituitary-dependent disease. Some centres recommend a single high dose overnight test of 8 mg or an intravenous test of 1 mg/hour for 7 hours. If there is good endocrine evidence of a pituitary ACTH secreting adenoma, transphenoidal surgery will result in cure in over 80 per cent and a low recurrence rate of 5 per cent. Plasma ACTH will be elevated, whereas it would be unmeasurable in adrenal disease. Hypokalaemia favours ectopic ACTH secretion. On the corticotrophin-releasing hormone test with bilateral inferior petrosal sampling an exaggerated serum cortisol response over basal suggests pituitary-dependent disease. Pituitary and adrenal MRI and chest imaging may identify an ectopic source. These tumours are frequently microadenomas and transphenoidal surgery is the treatment of choice. 15 to 40 per cent of patients with ACTH-secreting macroadenomas have a poor response to surgery and post-operative radiotherapy, and adrenal blocking drugs or bilateral adrenalectomy are commonly required. Patients who have received pituitary radiation therapy are at long-term risk of hypopituitarism and require continuing endocrine assessment (Brada et al. 1993).

Growth hormone secreting adenomas can be treated with somatostatin analogues or bromocriptine where surgery is contraindicated or unsuccessful (Barkan et al. 1988). Bromocriptine can also lower growth hormone levels in up to 75 per cent of cases, but normal growth hormone levels are only achieved in 10 to 20 per cent, treatment must be lifelong, and there may not be any shrinkage of macroadenomas. Surgery or radiotherapy is usually required. The growth hormone response to transphenoidal resection is not as good as the endocrine response to surgery for ACTH-secreting tumours. The aim is to achieve basal GH levels of <5 mu/l, suppressing to <2 mu/l after glucose loading. This will not be achieved in almost 50 per cent after trans-sphenoidal surgery. Post-operative radiotherapy is frequently required. Radiotherapy is moderately effective and may be used as the primary treatment where surgery is contraindicated or as an adjunct to surgery. It may take many years to achieve normal growth hormone levels and usually a

combination of surgery and radiotherapy or radiotherapy and drugs are required (Bloom et al. 1984).

Meningiomas are benign slow growing intracranial extracerebral tumours that account for 15 to 20 per cent of all intracranial tumours. Only 25 per cent of meningiomas are symptomatic at presentation. The frequency of meningiomas increases with age (Radhakrishnan et al. 1995). Women are more than twice as commonly affected. Risk factors include gender, previous ionising radiation, and neurofibromatosis type 2. The World Health Organization histological classification and grading system selects characteristics that predict an aggressive behaviour of the tumour and the risk of early recurrence (Kleihues et al. 1993b). There are many histological sub-types of meningioma and these tumours can be graded as benign meningiomas, atpical meningiomas or malignant meningiomas depending on the degree of anaplasia, cellular atypia, mitoses, and necrosis. Proliferation indices identified by immunohistochemical methods on pathological specimens may also predict an aggressive nature. Meningiomas express progesterone and oestrogen receptors and receptors for platelet derived growth factor (Black et al. 1996).

Eighty five per cent of meningiomas are supratentorial. The most common sites are over the convexities of the skull, the falx, or tentorium followed by the sphenoid ridge, suprasellar areas, and olfactory groove. CT and MRI will demonstrate a well-demarcated enhancing lesion with a dural base that may involve or displace adjacent nerves, or produce significant oedema in adjacent brain (Fig. 27.7). Angiography is not usually necessary; however, it may be helpful to identify the feeding vessels and allow immediate pre-operative embolization to reduce the potential for severe haemorrhage from the tumour during operation. If the pathologist is not informed that pre-operative embolization has been performed an incorrect diagnosis of malignant meningioma may be made because of the necrosis in the specimen. In symptomatic meningiomas with brain oedema, dexamethasone 2 to 4 mg t.i.d. will usually produce speedy relief of symptoms. Although most meningiomas are benign tumours, operation is not always straightforward. The extent of surgical excision is graded by Simpson’s grading of extent of surgical resection (Simpson 1957). Asymptomatic meningiomas, especially in the elderly, are best left alone. Large symptomatic meningiomas will usually require surgery, but pre-operative embolization may reduce the vascularity of the tumour and make surgery easier. Surgical mortality can be as high as 14 per cent and the 10 year survival can range from 43 to 77 per cent. Convexity, parasagittal, lateral sphenoid, and olfactory groove meningiomas can usually be resected completely with low morbidity. Suprasellar, cavernous sinus, clivus, tentorial, and posterior fossa meningiomas are more difficult, although improved surgical techniques have resulted in more radical resection. Morbidity is much higher in these sites and there can be a high recurrence rate. It has been estimated that the 10 year risk of recurrence is 9 to 20 per cent where the surgeon feels there has been a complete resection and 18 to 50 per cent recurrence where subtotal resection has been performed. Meningiomas at the base of the skull involving the sphenoid ridge may require a joint surgical approach by both facio-maxillary and neurosurgeons.

In symptomatic meningiomas in the elderly or at sites that increase operative risk, stereotactic radiation therapy or radiosurgery, as sole treatment, may reduce the size of the tumour or slow the growth rate. Radiosurgery is usually only considered for relatively small tumours of less than 3 cm in diameter that do not impinge on structures such as the pituitary gland or optic nerves, abut the ventricles, where operation would be hazardous or following subtotal resection. In these selected situations, 96 per cent of patients with benign meningiomas, 76 per cent with atypical meningiomas and 19 per cent with malignant meningiomas will have disease control at 5 years (Friedman et al. 2005). High dose fractionation regimes can be associated with 50 to 60 per cent short-term complications and 10 to 20 per cent permanent complications.

There is no role for hormonal therapy or chemotherapy in the management of benign meningioma. In recurrent, atypical or malignant meningioma, where repeated surgery is hazardous and maximal radiotherapy has been previously given, drug therapies have been tried. Antiprogesterones have been used with some apparent success in some patients with meningioma (Black 1993). Anti-oestrogens such as tamoxifen 40 mg/m² twice daily for 4 days followed by 10 mg twice daily, may produce a reduction in size of the tumour in 15 per cent of patients (Goodwin et al. 1995). Hydroxyurea has also been tried with limited success. In general, chemotherapies and not very effective and the balance of risk: benefit has not been proven.

Acoustic neuromas, schwannomas, account for approximately 80 per cent of extra-axial lesions in the region of the cerebellopontine angle and 4 to 10 per cent of intracranial tumours overall (Mahaley et al. 1990; Grant et al. 1996). Other tumours in the region of the cerebellopontine angle include meningiomas in 10 per cent, primary cholesteatoma in 5 to 10 per cent, glomus jugulare tumours in 1 per cent, facial or trigeminal neuroma in 1 to 2 per cent, and metastasis in 1 to 5 per cent. The incidence of acoustic neuroma is approximately 1/100 000/yr. The acoustic nerve is the most common site for a neuroma in 85 per cent, although they can also arise from the trigeminal nerve in 1 to 8 per cent, facial nerve in 0.5 to 1 per cent and spinal roots in 10 to 15 per cent. Schwannomas arise from the junction between the peripheral schwann cell nerve sheath and the central glial nerve sheath. Ninety five per cent of acoustic neuromas are sporadic and 5 per cent are dominantly inherited as part of neurofibromatosis type 1 or type 2. Karyotype analysis in sporadic schwannoma may be normal. The most common abnormality is monosomy of chromosome 22, and there may be deletions in the long arm of chromosome 22q. The NF2 gene has been isolated on chromosome 22q to a 6-Mb region of the q12 band of the long arm of chromosome 22 and it is highly likely that the relevant area on this chromosome involves a tumour suppressor gene. Mutations to the NF2 gene are also likely to be an important step in the pathogenesis of sporadic unilateral acoustic neuroma and have been found in 40–70 per cent of cases.

Up to 85 to 90 per cent of acoustic neuromas arise from the vestibular branch of the nerve. Acoustic neuromas present with unilateral slowly progressive hearing loss in 95 per cent, sometimes associated with non-specific unsteadiness in 77 per cent, tinnitus in 71 per cent or vertigo. Commonly the sensorineural deafness occurring in 90 per cent is associated with facial sensory loss in 50 per cent or facial weakness in 10 per cent. Hydrocephalus due to obstruction of CSF pathways can lead to raised intracranial pressure. MRI is the scanning procedure of choice.

The tumours are very slow growing. Almost half of the tumours do not grow perceptibly over a 5 year follow up. Of those that do enlarge, 75 to 80 per cent of cases the growth rate is only 1 to 2 mm/year (Bederson et al. 1991). There may therefore be a case for conservative management with careful follow-up rather than intervention, especially in the elderly or those in poor general health, where the tumour is small or if there is contralateral deafness and retained hearing in the affected ear. Growth rate apparently does not correlate with the age of the patient, the size of the tumour or the duration of symptoms (Bederson et al. 1991).

Surgical management has a high morbidity, particularly in patients with retained hearing. A suboccipital approach has the advantage of possibly retaining any existing hearing, but may require cerebellar traction and can cause post-operative headaches and cerebellar symptoms. The translabyrinthine approach has the advantage of requiring little in the way of cerebellar retraction, and because the surgery is largely extradural, the complications of meningitis and headache are less; however hearing is always lost post-operatively. The middle fossa approach is useful for small tumours and may spare hearing; however there are increased complication rates from facial nerve paresis and possibly sequelae such as seizures or dysphasia as a result of temporal lobe traction. Monitoring brainstem auditory evoked potential intra-operatively can significantly decrease the morbidity of surgery, especially when trying to preserve hearing. Prolongation of the latency of wave V of the brainstem auditory evoked potential is usually an early sign that the acoustic nerve is being compromised. Acoustic neuromas may require a joint surgical approach by both ear, nose, and throat surgeons and neurosurgeons. Surgery should aim to remove the tumour completely and to preserve facial nerve function and where possible preserve hearing. Neurosurgery for acoustic neuroma has a post-operative mortality of approximately 5 per cent. Mortality is related to the size of the tumour and age of patient (Hardy et al. 1989). These percentages are heavily influenced by selection of patients, experience of the surgeon, and possibly the surgical approach. Mortality using the translabyrinthine approach is probably not significantly different from the suboccipital or middle fossa approaches when patient characteristics are taken into account. There is general consensus that small tumours are better approached by a translabyrinthine approach whereas large tumours are best approached suboccipitally or by middle fossa approach. Translabyrinthine surgery results in complete hearing loss but this is not a problem if hearing is already lost pre-operatively and there may be slightly more chance of preserving facial nerve function. Anatomical preservation of the facial nerve can be achieved in about 80 to 90 per cent of cases, but anatomical preservation is not always associated with good facial nerve function. It is exceptionally rare for post-operative hearing to be better than pre-operative hearing using either the suboccipital or middle fossa approaches. Where hearing preservation is the main aim, for instance if there is already contralateral deafness and the affected ear has maintained hearing, a suboccipital approach has advantages. In selected centres it has been suggested that with experienced neurosurgeons and the aid of brainstem auditory evoked potentials, complete resection of the tumour can be accomplished with preservation of hearing in 50 per cent of patients with tumours smaller than 2 cm and in >80 per cent of patients who have a tumour of <1cm diameter (Post et al. 1995). However, a recent review shows that only a few patients have truly normal hearing after surgery (Sanna et al. 1995). Delayed deterioration in hearing years after successful operation is well recognized in up to 50 per cent of patients, although the cause remains uncertain (Shelton et al. 1990; Ogunrinde et al. 1994). Attempts at maintaining hearing by minimizing resection can be complicated by recurrence of the tumour.

In some cases where hearing is preserved and the tumour is small, stereotactic radiosurgery or stereotactic radiotherapy using a linear accelerator can be effective. Stereotactic radiosurgery uses a single fraction of high dose but small volume radiation to the tumour. Conformal beam stereotactic radiotherapy using a linear accelerator and fractionating the treatment over several days and reducing the dose of each fraction, has potential advantages in that radiation-induced neural side effects are reduced by reducing the fraction size. Radiation therapy is not usually advised for tumours of >3 cm because of the increased risks of central nervous system side effects. The aims of radiation therapy are to prevent growth of acoustic neuroma and retain neurological function. Tumours <3 cm in diameter show shrinkage or ‘stabilization’ in 97 per cent at 5 years after stereotactic radiosurgery (Flickinger et al. 2001). This apparent success has to be compared with the natural history of acoustic neuroma. One study of conservative management demonstrated that 71 per cent of acoustic neuromas do not enlarge over 3.4 years (Deen et al. 1996). Short-term complications from stereotactic radiosurgery or stereotactic radiotherapy using a linear accelerator are infrequent; however it will be some years before one can fully ascertain the effect of radiation on the acoustic nerve and surrounding structures, particularly in patients with normal pre-radiation hearing. Hearing is very likely to become impaired with time and only 50 per cent of patients with preserved hearing following radiation therapy will maintain this at 6 months and only 45 per cent at 1 to 2 years. Two years after stereotactic radiosurgery, preserved facial nerve function is achieved in 90 per cent and trigeminal nerve function in 75 per cent of cases who had no deficit immediately post-radiotherapy. In the long term, cases of radiation related cancer in the treatment field have been described. The results of conformal beam stereotactic radiotherapy are as good as stereotactic radiosurgery but longer term side effects appear to be less, probably reflecting the fractionation schedule and reduction in fraction size. Where surgery is contra-indicated because of poor health or poor risk: benefit ratio, radiosurgery or conformal stereotactic radiotherapy should be considered the treatment of choice, and there is adequate evidence that these treatments have a therapeutic role. The role of radiotherapy in the treatment of small acoustic neuromas remains controversial.

Cerebral gliomas are the most common primary intrinsic brain tumours. Gliomas are locally invasive and even after apparently successful macroscopic resection they recur at the same site in 95 per cent of cases. They spread outwith the central nervous system in less than 1 per cent, although it has been estimated that CSF spread occurs in up to 5 per cent.

For practical purposes gliomas can be divided into low grade gliomas, World Health Organization or Dumas–Duport grade 1 and 2, and high grade gliomas, World Health Organization or Dumas–Duport grade 3 and 4.

Low grade gliomas account for approximately 20 per cent of all cerebral gliomas. Symptoms will depend on the site of the tumour. Prognosis depends on age at diagnosis, length of pre-operative symptoms, epilepsy, and extent of resection (Piepmeier et al. 1996; Salcman 1995). World Health Organization grade 1 gliomas include rare entities like pilocytic astrocytoma, pleomorphic xanthoastrocytoma, and subependymal giant cell astrocytoma and are potentially curable if they can be completely resected. Post-operative radiotherapy is not required. If there is suboptimal resection, these tumours are so slow growing that it is unlikely that early cranial radiotherapy has any advantage over a wait and watch policy, particularly since radiation induced side effects increase with the passage of time and there can be extended survival even in the group with subtotal resection. In symptomatic cases where the risks of resective surgery are considered too great, radiation therapy can produce long-term symptomatic benefit. Approximately 80 per cent of patients with pilocytic astrocytomas are alive at 15 years (Shaw et al. 1997; Shaw 1995).

World Health Organization grade 2 gliomas, fibrillary or protplasmic astrocytomas, oligoastrocytomas, and oligodendrogliomas, commonly present with seizures without neurological deficit. In adult patients with low grade glioma, older age, astrocytoma histology, presence of neurologic deficits before surgery, largest tumour diameter > 6 cm and tumour crossing the midline were important adverse prognostic factors for survival. These factors can be used to identify low-risk and high-risk patients (Pignatti et al. 2002). Patients under the age of 40 years have a median survival of 8 years compared with 5.5 years for patients aged between 40 and 50 years and 1.6 years if older than 50 years. If patients have a good performance status, the median survival is 7.4 years compared with 1.6 years if they have a poor performance status (Eyre et al. 1993). Most tumours are situated in the frontotemporal regions and are frequently diffuse, extending throughout a lobe at presentation (Figs 27.8, 27.9, and 27.10). Clearly the diffuse infiltrating astrocytomas cannot be resected. For more focal low grade astrocytomas in non-eloquent areas, resection may be feasible. There is uncertainty of how best to treat patients with low grade gliomas. It is probably advisable to biopsy lesions suggestive of low grade glioma because there can be foci of higher grade despite the lack of enhancement. However, if seizures are the only symptom, a wait and watch policy is preferred by some clinicians and patients. The growth rate of low grade gliomas is very slow and has been estimated as 4.1 mm/year (Fig. 27.11). There are no randomized controlled trials of resection versus biopsy in low grade glioma and it is unlikely that such a study would see completion because of the excellent survival with astrocytomas, with 46 per cent five year survival and oligodendroglioma 73 per cent five year survival (Shaw et al. 1997; Shaw 1995). Radiation can reduce the size of a tumour but because of the good survival, this has to be balanced with the increased likelihood of developing delayed radiation-induced toxicity. A randomized controlled trial of early versus delayed radiotherapy for low grade gliomas accrued 314 patients. Median progression free survival was 5.3 years with early radiotherapy compared to 3.4 years for the wait-and-see group; however, overall survival was similar in both groups at 7.4 years compared to 7.2 years respectively. In the control group, 65 per cent received radiotherapy at progression. The study did not examine late complications of radiotherapy or quality of life (Karim et al. 2002; van den Bent 2005). A randomized study of radiation versus radiation plus Choroethyl-Cyclohexyl-Nitrosourea, CCNU, demonstrated more radiological responses in the radiation alone group, at 79 per cent versus 54 per cent, failed to demonstrate any survival advantage with adjuvant chemotherapy and there were significant haematological toxicities in the chemotherapy treated group (Eyre et al. 1993). More recently, PCV chemotherapy with Procarbazine, CCNU, and Vincristine or Temozolomide have been advocated prior to radiation therapy as this can occasionally lead to reduction in tumour size or symptomatic improvement in low grade glioma; however, the role of chemotherapy has yet to be proven. At the time of recurrence most tumours have changed to higher grades of malignancy and chemotherapy may have something to offer (Muller et al. 1977).

High grade gliomas: anaplastic astrocytoma, glioblastoma multiforme, and anaplastic oligodendroglioma, account for

approximately 80 per cent of all cerebral gliomas. Prognosis depends on age, grade, performance status, and possibly, the extent of remaining disease after surgery. Age is the most important independent prognostic variable at presentation followed by grade. A recursive partitioning analysis is available from three Radiation Therapy Oncology Group trials (Curran et al. 1993). Patients with poor performance status are less likely to be offered treatment and the patients with poor performance status invariably also have shorter survival, even accounting for age and grade. Site of tumour and volume of tumour on pre-operative CT or MRI does not seem to be prognostically important. Some studies suggest that the amount of peritumoural oedema, extensive contrast enhancement or the volume of tumour remaining on a scan performed at 48 to 72 hours post-operatively are associated with poor outcome (Hammoud et al. 1996; Muller et al. 1977; Piepmeier et al. 1996). Care must be taken when interpreting enhanced images with respect to timing of injection of contrast and the time of performance of the scan. Tumours can significantly alter in their enhancement on dynamic scanning depending on the delay between contrast injection and performance of imaging (Fig. 27.12).

It is not certain whether the type of surgery performed, such as biopsy versus resection, has any survival benefit other than in patients who are on the verge of coning or who have hydrocephalus. Sterotactic biopsy can undergrade a glioma, because of inadequate sampling which in turn may result in patients with stereotactic biopsy anaplastic astrocytomas appearing to do worse than those where the diagnosis has been based on an extensive resection, purely because of grading error (Glantz et al. 1991). One or two prospective studies have collected data before or after resection and show that about one third improve after resection, 58 to 76 per cent stay the same and 9 to 26 per cent deteriorate (Fadul 1988; Sawaya 1998). There are several phase 2 studies and a recursive partitioning analysis that support the view that extent of resection influences survival (Simpson et al. 1993; Schiff and Shaffrey 2003). These however are affected by a number of confounding variables, including size, site, age, further treatment, and lack of a comparable control group, and it is not possible to say with certainty that resective surgery extends median survival. A Cochrane systematic review (Hart et al. 2007) identified only one randomized controlled trial of biopsy versus resection in patients with radiologically ‘obvious’ glioblastoma multiforme in patients older than 65 years (Vuorinen et al. 2003). This study of 30 patients randomized to either maximum resection or stereotactic biopsy reported that 23 per cent had diagnoses other than glioblastoma, consisting of stroke, metastasis, lymphoma, or uncertainty, leaving 13 in the biopsy group and 10 in the resection group who were referred for radiotherapy. Data was available in 18 of these patients; the overall median survival time was 21 weeks, with resection survival of 24 weeks versus biopsy of 12 weeks (p <0.035). There was no significant difference in the time to deterioration between the groups; the amount of radiotherapy received had an effect on survival (p <0.001) (Vuorinin et al. 2003). This study was underpowered to give any firm conclusion to whether resection prolongs survival.

If biopsy is done this should be by stereotactic technique rather than freehand, because of the higher complication rate associated with the latter. New operative adjuncts, including awake craniotomy, functional MRI, and intra-operative MRI have been used in selected cases in highly specialized centres where the tumour is in an eloquent area. One small randomized controlled trial of neuronavigation did not show any survival advantage or reduction in death or disability (Willems et al. 2006). The goal of resective neurosurgery should be as complete resection as possible along its macroscopic boundaries. If achieved without complications, this provides reliable histological diagnosis, potentially improves the patient’s neurological status and may make the tumour more sensitive for additional therapies, such as chemotherapy (Salcman 1987; Shapiro et al. 1989). The degree of tumour removal in most studies has been determined by the intra-operative perception of the neurosurgeon. With the increasing availability of neuro-imaging, it has become clear that the surgeon’s opinion at the time of operation of what represents a total resection bears little resemblance to the post-operative MRI appearances. Intra-operative imaging is only performed in a few centres. 5-Aminolevulinic acid, Gliolan^™, is a chemical that is taken up preferentially by brain tumour tissue and fluoresces under blue light. It has been used to assist the surgeon in identifying remaining tumour at surgery and so may help improve the extent of resection of gliomas. A randomized controlled trial in selected patients with peripherally situated gliomas which demonstrated that it was possible to resect more, it took longer for the tumour to return on imaging, but it did not affect the overall survival. There was a slightly higher early neurological complication rate in the group given 5-Amino-levaliric acid (Stummer et al. 2006).

Focal complications following surgery include haematoma, abscess, stroke, and seizures, whilst systemic complications include thromboembolism and pneumonia. Morbidity rates range from 11 to 32 per cent with post-mortality rates ranging from 0 to 20 per cent. Variations in reporting these complications and how long after surgery the mortality is recorded will account for some of the larger variations. The complication rates are lower for biopsy than for resection. Post-operative enhancement on CT scan performed before the 5th post-operative day reflects residual tumour (Cairncross et al. 1985; Jeffries et al. 1981). Examination of serial post-operative MR scans has demonstrated that post-operative imaging during days 1 to 3 after resection of a high grade glioma avoids artefacts due to post-operative enhancement and the delineation of tumour was vastly superior to post-operative CT (Albert et al. 1994). MRI studies have suggested that post-operative residual tumour was a more important prognostic variable than age or performance status and the incidence of tumour recurrence is directly related to the volume of residual tumour after initial resection (Albert et al. 1994; Berger 1995). However, these results must be interpreted with caution since the selection of patients for aggressive resection based on tumour location and demarcation from surrounding normal tissue may introduce selection bias.

Malignant glioma is one of the most aggressive tumours in man. It rarely spreads outwith the central nervous system, is highly radioresistant, and has a predilection to loco-regional recurrence. Each of these three factors has lead to particular approaches to primary treatment and management of ‘recurrence’. There is good randomized controlled evidence from the early 1970s and 1980s that radiation therapy improves survival in patient with high grade gliomas (Walker et al. 1978, 1980). Radiation therapy increases the median survival from 4 to 5 months to about 9 months. A randomized controlled trial has demonstrated that 60 Gy in 30 fractions over 6 weeks was superior to 45 Gy in 20 fractions over 4 weeks and resulted in a prolongation of survival by 3 months in the group treated with 60 Gy (Bleehen et al. 1991). The current standard practice is to give 60 Gy in 30 fractions over 6 weeks. Radiation therapy is usually directed at the area of the enhancing tumour plus at least 2 cm margin of peritumoural oedema (Chang et al. 1983; Halperin et al. 1996). Focal radiation with 40 Gy in 20 fractions over 4 weeks to the tumour and peritumoural oedema is usually followed by a further 20 Gy boost to the enhancing tumour and 1 to 2 cm margin over 2 weeks. The wide margins are because tumour cells can be found 2 cm or more from the apparent radiological boundary of the tumour in areas that simply look ‘oedematous’ on CT or MRI and most studies demonstrate that relapse occurs within 2 cm of the enhancing rim of the tumour in 80 per cent of cases (Halperin et al. 1989; Wallner et al. 1989). Boosting the radiation dose to the centre of the tumour is now standard practice in most centres. Dose escalation of radiation using conformal beam therapy or stereotactic radiation aims to maximally treat the centre of the tumour and spare normal tissue outwith the 2 cm margin to reduce long-term morbidity from radiation damage.

Hyperfractionation, multiple small fractions per day to a higher overall dose, and ‘acceleration’ consisting of multiple treatments per day with the same dose but reducing the overall treatment time, do not improve survival. Focal therapies including interstitial brachytherapy, neutron therapy, and particle pions have not revealed a significant survival advantage in small randomized controlled trials (Laperriere et al. 1998; Duncan et al. 1986; Pickles et al. 1997). The effect of radiation is greater in those under 60 years. In the over 70 age group the effect of radiation therapy remains controversial, but one recently reported randomized controlled trial in patients >70 years of age with anaplastic gliomas or glioblastoma who had a Karnofsky performance score of >70 found a 8 week improvement in progression free survival and 11 weeks median overall survival from standard radiotherapy consisting of 50 Gy in 1.8 Gy fractions, when compared with standard care (Keime-Guibert et al. 2005). The 1 year survival was 13.8 per cent with radiotherapy compared with 2.6 per cent in those managed conservatively. A Phase III study that compared the addition or stereotactic radiotherapy to conventional radiotherapy plus BCNU with conventional radiotherapy plus BCNU alone for supratentorial malignant glioma failed to demonstrate any survival advantage or changes in patterns of failure with the addition of stereotactic radiotherapy (Souhami et al. 2004).

Numerous chemotherapy agents, such as Nitrosoureas, Procarbazine, or Platinum derivatives, have been shown to reduce tumour size in about one-third of patients in phase II studies (Mahaley 1991). The beneficial effect of chemotherapy has to be balanced with potentially serious side effects. The risk: benefit ratio will depend on individual patient- and tumour-related factors. Younger patients, patients with anaplastic astrocytoma of grade 3, and patients with pure or mixed oligodendroglioma, are more likely to harbour chemosensitive tumours. Tumour-related genotypic factors such as chromosomal deletions on chromosomes 1p and 19q in oligodendroglioma and DNA repair enzyme inactivation by methylation at the promotor site of Methylguanine-DNA methyltransferase, MGMT, in glioblastoma. Approximately 50 per cent of malignant gliomas have an inactivated MGMT and these patients are associated with better prognosis and response to treatment with alkylating agents such as BCNU or Temozolomide, whereas tumours with unmethylated MGMT have a poorer prognosis and are very unlikely to respond to chemotherapy (Hegi et al. 2005).

When considering the lack of long-term efficacy from chemotherapy, and methods to overcome this, one has to examine primary mechanisms for drug resistance. Firstly, it must be possible to deliver the drug to the site of the tumour in sufficient concentration to have an effect. Secondly, drug must be able to pass through the blood brain barrier at the ‘advancing’ edge of the tumour, which may not be well vascularized. Chemotherapy penetration at the necrotic centre or enhancing nidus, will not prevent tumour progression at the margin. Thirdly, intracellular mechanisms of drug resistance, such as lack of intracellular drug activation, drug inactivation for instance by glutathione-s-transferases, DNA enzyme repair for instance by O⁶-alkylguanine-DNA-alkyltransferase and active removal of chemotherapy from the cell by P-glycoprotein, an energy-dependent drug efflux pump mediated through multidrug resistance gene overexpression, must be overcome. Increasing drug delivery by intra-arterial chemotherapy, can reduce systemic effects of chemotherapy while maintaining the local concentration of chemotherapy to the tumour, however, the potential benefits are limited by local toxicity to the brain with encephalopathy or seizures and to the eye causing optic neuropathy. Randomized studies have shown no benefit in terms of survival over systemic chemotherapy (Mahaley et al. 1986; Green et al. 1989). Chemotherapy with BCNU and Cisplatin given prior to radiation was found to be no better than adjuvant chemotherapy and had more toxic side effects (Grossman et al. 2003). Most studies therefore seem to support the view that chemotherapy can be useful whether given adjuvantly or at the time of clinical or radiological relapse after radiation therapy, although no one has achieved the 25 per cent 2 year survival achieved with concomitant and adjuvant Temozolomide compared with only 10 per cent in the control arm (Walker et al. 1978; EORTC Brain Tumor Group 1981; Stupp et al. 2005). This in part may be due to more stringent entry criteria. Temozolomide is less effective in biopsied cases and cases with moderate disability, a World Health Organization performance status of 2 or above. The long-term complication rate, in terms of radiation-induced leucoencephalopathy, of the radiation therapy plus concomitant and adjuvantly Temozolomide group remains uncertain.

Gene therapy techniques are currently being used to try and target elements in the tumour gene. Many genes have been implicated in the development and growth of gliomas. Despite very promising phase II studies and tremendous enthusiasm, a phase III trial of gene therapy with Herpes Simplex Virus-tk and radiation therapy did not show any advantage over radiation therapy alone (Rainov 2000). The gene therapy was injected directly into the tumour resection boundaries after maximal tumour resection and this was followed by radiotherapy. The failure may have in part been due to a very poor rate of transfection. There are other methods using adenoviral vectors with improved gene transfer efficiency of up to 10 per cent and stronger transgene expression in the tumour, which in one small pilot randomized studies suggest a benefit (Immonen et al. 2004). This study however was underpowered and the active treatment group had a lower median age and fewer glioblastoma multiforme patients.

The main emerging approaches to treatment of malignant glioma involve ‘Targeted therapies’. These treatments target genetic and cellular alterations in cell signalling that are more commonly found in tumour cells such as amplification of oncogenes such as epidermal growth factor receptors on the cell surface or mutations of tumour suppressor genes, for instance, the phosphatase and tensin homologue, PTEN. Blood vessels can also act as a target such as vascular endothelial growth factor, VEGF, and platelet derived growth factor, PDGF, which are over-expressed. EGFR, VEGF, and PDGF also activate pathways which result in up-regulation of the anti-apoptotic protein Bcl-xl which help cells survive. These factors lead to proliferation of tumour cells, stimulate angiogenesis, and reduce the ability for cells to become apoptotic. Proliferating cells probably express different receptors to invading tumour cells. Cells in the process of invading over-express integrin cell surface adhesion molecules and matrix metalloproteinases that degrade extracellular proteins and allow the tumour cells to migrate and invade into surrounding brain.

There are no tumour specific antigens, but there are tumour associated antigens. These can act as targets for antibodies which may down-regulate or stop the action of oncogenes or growth factors, or be used to attach conjugated toxins, radiotherapy, or chemotherapy. These tumour associated antigens can be targeted by monoclonal antibodies such as Cetuximab which can block EGFR; EGFR tyrosine kinase inhibitors such as gefitinib, erlotinib, or imatinib; conjugated toxins such as pseudomonas exotoxins conjugated to transforming growth factor alpha or interleukin-13; anti-angiogenic agents, such as thalidomide and anti-invasive agents such as Marimistat or cilengitide. There are also some novel cell growth and migration inhibitors such as Accutane. These all act at a molecular or biological target level. Cetuximab binds to EGFR and prevents ligand binding, blocks ligand-induced tyrosine kinase activation and receptors become internalized. It induces apoptosis and inhibits angiogenesis in vitro and has shown some success in phase II clinical trials. I¹³¹ labelled murine anti-tenascin monoclonal antibody 81C6, Neuradiab^™ injected directly into the tumour rim after resection appears to improve survival in phase II studies when compared with historical controls. Monoclonal antibodies conjugated with isotopes such as Yttrium or pseudomonas exotoxin (Precise) have also been used in small phase II studies and are currently in phase III studies. Transferrin receptors are expressed tumour associated antigens and transferring conjugated with diphtheria toxin, Transmid^™, is in phase III studies. Tyrosine kinase inhibitors such as erlotinib have been used in malignant glioma and have produced a 20 per cent response in selected groups. These small molecules are now being increasingly used in combination with standard therapies such as radiotherapy and chemotherapy. Because these small molecules are likely to be broken down systemically, most studies involve direct injection at the time of surgical resection or infusions into the tumour cavity using convection enhanced delivery technologies within the sealed tumour cavity given under pressure to allow diffusion of the small molecule to 2 cm from the cavity. Thalidomide inhibits integrin receptors and inhibits neovascularization. It is a drug notorious for its teratogenic effects in the foetus causing in particular phocomelia, and is contraindicated in women of childbearing age who have a chance of pregnancy. Matrix metalloproeinases such as Marimistat have been used in malignant glioma along with Temozolomide, but can cause joint pain in up to 50 per cent.

New small molecule therapies may be associated with hitherto unexpected side effects. Many treatments will only be possible if the patient is fit enough for resection. They may only be associated with inhibition of tumour progression, rather than causing tumour regression. New protocol designs and clinical response measures may be necessary and the likely way forward will be combinations of these therapies accompanied by standard treatments. The treatments are expensive. It seems unlikely that one particular small molecule or single target will lead to cure since only a percentage of tumours express these antigens and not all tumour cells express a specific target. They need to be proven in large well designed randomized controlled trials.

Although malignant glioma continues to have a poor prognosis, anaplastic oligodendroglioma can be responsive to chemotherapy for prolonged periods, and chemotherapy prior to radiation therapy may prove to be more effective. Studies suggest that PCV chemotherapy produces a 75 per cent response rate, with 17 per cent stable disease and only 8 per cent have progressive disease (Macdonald et al. 1990; Kritis et al. 1993; Macdonald 1994). In this moderately chemoresponsive tumour, where a large proportion of patients will respond, even those >60 years, there is some evidence that patients who have a ‘major response’ defined as >90 per cent reduction in contrast enhancing area, have better survival than those with partial response or stable disease (Cairncross and Eisenhauer 1995; Macdonald 1994). As mentioned above, tumours with Ch 1p and Ch 19q deletions seem to do particularly well with chemotherapy. A prelimary report of a randomized controlled trial of pre-radiotherapy PCV versus radiotherapy followed by chemotherapy as necessary demonstrated that pre-radiotherapy PCV does not impart a survival advantage for histologically defined anaplastic oligodendrogliomas and anaplastic oligoastrocytomas, but may prolong progression-free survival at the expense of greater acute toxicity (Cairncross et al. 2004).

Medulloblastoma is the most common childhood central nervous system tumour, but is a relatively rare tumour in adults. About 25 per cent of all medulloblastomas occur in adults, aged >16 years. The cell of origin of this tumour remains uncertain but is probably of embryonic origin with the possibility of taking different lines of differentiation, otherwise known as primitive neuroectodermal tumour or PNET. The most common genotypic abnormality, in 30–45 per cent of cases, is the loss of genetic material from chromosomal arm 17p (Ellison 2002). This site is apparently the location of a suppressor gene, the removal of which allows for the expression of the tumour. Medulloblastomas usually arise in the midline and are most commonly found in the cerebellum, approximately 95 per cent in children. Presentation is usually either related to a cerebellar ataxia, or raised intracranial pressure related to obstructive hydrocephalus. Cranial nerve presentations, especially diplopia, also occur. The tumour has low signal on T1-weighted MRI and high signal on T2-weighted scans and there is usually gadolinium enhancement on T1-weighted images.

Primitive neuroectodermal tumours are malignant tumours with a high propensity to disseminate throughout the CSF. Staging of the tumour is therefore important, but can wait till after the definitive surgical procedure. Because of the possible contamination of CSF at the time of surgery it is probably best to defer CSF analysis till 10 to 14 days after the operation; however the MRI of the spine to look for gadolinium enhancing nodules can be performed any time after surgery. Thirty to forty per cent of children will have CSF dissemination although this may be greater in children under 5 years of age (Deutsch 1988). The bad prognostic features for medulloblastoma appear to be age <3 years, CSF dissemination, and possibly extent of resection (Evans et al. 1990). Current treatment guidelines are largely determined by clinically based prognostic factors, the most important of which are tumour location and the extent of tumour spread. Although the cure rate for high-risk primitive neuroectodermal tumours has improved, the irreversible sequelae of craniospinal axis radiation treatment in patients who survive have motivated researchers to investigate more fully which patients can safely receive less treatment (Jakacki 2005).

The management of acute obstructive hydrocephalus is controversial. Some surgeons suggest steroids and tumour resection, whereas others suggest steroids and external ventricular drainage or steroids and ventriculoperitoneal shunting prior to a definitive operation. Surgery for medulloblastoma is performed in the prone position, to avoid the risks of air embolus and pneumocephalus or systemic hypotension when operating in the seated position. A repeat MRI scan at 48 hours may be helpful for assessing the extent of tumour resection.

Radiotherapy to the tumour improves survival in children with medulloblastoma. In addition there is good evidence that craniospinal irradiation reduces the risk of recurrence from CSF dissemination. A randomized prospective trial demonstrated improved disease control using 36 Gy of craniospinal irradiation with a posterior fossa boost to 54 Gy compared with 23.4 Gy with a posterior fossa boost to 54 Gy in 30 fractions (Deutsch et al. 1996). This study was prematurely closed after an interim analysis at 16 months demonstrated a significant number of relapses in the low dose craniospinal irradiation group. Craniospinal irradiation also improves the number of 10 year survivors (Landverg et al. 1980; Castro-Vita et al. 1980). Radiation therapy to the craniospinal axis can have serious long-term toxicities: neuropsychological, neuro-endocrine with growth retardation and hypothyroidism, bone marrow depression, and second neoplasms in long-term survivors particularly of thyroid and other central nervous system neoplasms. In attempts to try and reduce these toxicities and to try to improve survival, chemotherapy has been used. Medulloblastoma is a relatively chemosensitive tumour. Randomized studies have compared radiation with radiation and CCNU, vincristine chemotherapy. The overall survival was 53 per cent at 5 years and 45 per cent at 10 years. Although an early analysis demonstrated a significantly better disease free survival with the addition of chemotherapy, there were subsequently late relapses in the chemotherapy group and no statistically significant effect on disease free survival was apparent. On post-hoc subgroup analyses, chemotherapy was thought to benefit patients with subtotal resection, brainstem involvement, and more extensive disease at presentation (Tait et al. 1990). A further randomized controlled study of the same agents in North America came to the same conclusions but also noted a survival advantage in patients with extensive disease in a subgroup analysis (Evans et al. 1990).

Currently, there is a vogue for chemotherapy in addition to craniospinal radiotherapy with a boost to the primary site for patients with a high risk of progression, in an attempt to improve survival and reduce the dose of craniospinal irradiation (Cohen et al. 1996). Approximately 10 to 20 per cent of patients will eventually have metastatic spread outwith the central nervous system, especially to lungs and bone.

With treatment, approximately 33 to 66 per cent of children are alive at 5 years and 25 to 50 per cent are alive at 10 years, depending on the patient selection for the survival analyses (Bloom et al. 1991). Younger age at diagnosis was the most prominent risk factor for neurocognitive deficits among survivors, despite reductions in craniospinal irradiation dosing and efforts to limit the boost volume. Young patients show substantial problems with the development of reading skills and intelligence (Mulhern et al. 2005).

Primary central nervous system lymphoma accounts for approximately 1 per cent of all primary intracerebral tumours and is almost always B-cell in origin. They occur in higher frequencies in patients with some form of immunosuppression. Approximately 2 to 6 per cent of patients with AIDS and 0.5 to 1 per cent of patients following transplantation will develop a primary central nervous system lymphoma. Primary central nervous system lymphoma is also found in association with Wiskott Aldrich syndrome, systemic lupus, idiopathic thrombocytopenic purpura, Sjogren’s, and sarcoidosis (Ling et al. 1997; Remick et al. 1990). Despite the association with immunosuppressive states, primary central nervous system lymphoma more commonly occurs sporadically in the immunocompetent. Epstein–Barr virus can be detected in all AIDS-related primary central nervous system lymphomas raising the possibility of a direct virus related cause (MacMahon et al. 1991; Guterman et al. 1996). The incidence was increasing in the 1990s but has stabilized or may be falling (Kadan-Lottick et al. 2002). Primary central nervous system lymphoma is classified as an extranodal lymphoma, stage 1_E, and it most commonly presents as single or multiple contrast enhancing space occupying lesion(s) within the brain or as cells in the eye, ‘vitreous lymphoma’, or CSF, ‘leptomeningeal lymphoma’. Primary central nervous system lymphoma is most commonly confined to central nervous system or eye and even at post-mortem only 10 per cent of patients are found to have disease outwith the central nervous system. The tumour is commonly in a periventricular site and this may explain the high incidence of CSF involvement. Approximately 20 per cent of patients will have CSF involvement at the time of presentation and a further 20 per cent will have had or develop a uveitis with ‘floaters’ and progressive loss of vision due to lymphoma of the vitreous (Hochberg et al. 1988). The diagnosis may be suspected if the CT or MR scan is typical, but biopsy is essential. There is no evidence that resection is superior to stereotactic biopsy. The tumours frequently lose their enhancement when steroids are given and it is not uncommon for the tumour to ‘disappear’ on CT scan after steroids which makes stereotactic biopsy difficult (Fig. 27.14). These tumours have an increased propensity to bleed when biopsied compared with other intracerebral tumours. Staging of primary central nervous system lymphoma is important. The International Primary Central Nervous System Lymphoma Collaborative Group has published guidelines for standardized baseline evaluation in newly diagnosed cases (Abrey et al. 2005). These include centralized pathology review and immunotyping; neurological, ophthalmological, slit lamp, and cognitive examination; HIV serology, lactate dehydrogenase, CSF cytology, flow cytometry, and 24 hour urine for creatinine clearance; contrast enhanced MRI of brain, CT scan of chest and abdomen, bone marrow biopsy, and testicular ultrasound in elderly males. Slit lamp examination may identify uveitis due to vitreous involvement, retinal detachment or haemorrhages, optic neuropathy or papilloedema. Pathologists may have difficulty identifying lymphomatous cells from chronic inflammatory cells and both populations can co-exist. The CSF protein is usually increased and one third have CSF glucose levels below the lower limit of the accepted range. CT or MR scan of the chest and abdomen and bone marrow biopsy is almost always negative (Hochberg and Miller 1988).

Corticosteroids are cytotoxic to lymphocytes and up to 40 per cent show a significant initial response, and occasional prolonged remissions with steroids have been reported. Patients younger than 60 years treated with >40 Gy to the whole brain and a boost to 50 Gy to the tumour bed along with chemotherapy appear to have a better survival with a median of around 2 to 3 years (Reni et al. 1997). Escalating the boost to 60 Gy does not appear to produce any survival advantage (Nelson et al. 1992). Chemotherapy used to be kept in reserve for relapsed primary central nervous system lymphoma; however, now chemotherapy with high dose methotrexate has been found to be effective and has good CSF penetration (Lachance et al. 1994; Glass et al. 1994), the trend is to give chemotherapy and radiotherapy in patients <60 years. With the combined regimes 50 per cent respond and 2 year survival rates in up to 70 per cent of selected patients (Ferreri et al. 2003). Some authorities suggest chemotherapy only in patients >60 years, because of high frequency of neurotoxicity with radiation and chemotherapy (Batchelor et al. 2003). Others consider that chemotherapy-related toxicity in the elderly is so frequent and prolonged remissions so few that treatment with chemotherapy is not justified, especially in patients with other co-morbidities. These various protocols have not been compared in randomized controlled trials.

Patients older than 60 years given chemo-radiation have a very high incidence of treatment-related neurotoxicity with dementia, ataxia, and incontinence with a median time to onset of about a year (Abrey et al. 1998). Neurotoxicity is much less common in patients under 60 years with a frequency of 30 per cent during 96 months of follow-up. Intrathecal methotrexate used to be the treatment of choice for CSF disease, but this treatment is no longer required if high doses intravenous methotrexate are given. Ocular disease can be treated symptomatically by steroids and frequently is misdiagnosed as non-lymphomatous related uveitis. Steroids usually only have a temporary effect and radiation to the posterior two-thirds of the eye or chemotherapy with high dose cytosine arabinoside or methotrexate can be useful.

Radiation therapy plus or minus chemotherapy certainly produces a clinical and radiological response in most cases and the median survival is in the region of 2 to 3 years in non-immunocompromised patients. The median survival of AIDS patients with primary central nervous system lymphoma, treated with radiation plus or minus chemotherapy, is approximately 3 months.

Pineal region tumours can arise from a number of cell types. The most common tumours in this area are teratomas or germinomas. Pineocytomas are less common and arise from the pineal parenchymal cells and gliomas can also arise in the pineal region. Occasionally metastases can spread to the pineal region. In addition to these solid tumours, pineal region cysts can occur and can be simple, filled with CSF or can be epidermoid or dermoid cysts. It can be difficult to distinguish with any certainty by imaging characteristics whether a pineal region tumour is a pineal glioma, an ependymoma, a pineocytoma, germinoma, or teratoma. Serum and CSF analysis can be helpful. If germ cell markers such as alphafetoprotein and beta human chorionic gonadotropin are positive, biopsy of the lesion is not required. Elevated germ cell markers indicate a malignant germ cell tumour. These patients can be treated with radiation therapy and chemotherapy and followed by measuring the tumour markers. If the lesion enlarges but markers reduce, surgical debulking may be necessary (Lee et al. 1995). In the absence of tumour markers, surgical confirmation is strongly advised. Hydrocephalus may have to be dealt with first, either by external drainage or placement of a ventriculo-peritoneal shunt. In the presence of mild hydrocephalus, some surgeons prefer performing a definitive resection and placing a ventricular drain at the time of surgery, which can be later clamped and removed or changed into a ventriculo-peritoneal shunt. Other authors have suggested performing a third ventricular ventriculostomy (Goodman 1993).

If the frozen section reveals germinoma, then there is no need to proceed to complete resection because this tumour is so radiosensitive and chemosensitive. Cure rates for germinoma exceed 90 per cent at 10 years. If the biopsy confirms a benign pineal region tumour, such as dermoid, epidermoid, pilocytic astrocytoma, ependymoma, or pineocytoma, maximum resection should be attempted and is generally the only treatment necessary. If the biopsy demonstrates a malignant pineal region tumour, such as a non-germinomatous germ cell tumour, choriocarcinoma, embryonal cell carcinoma, immature teratoma, endodermal sinus tumour, or malignant glioma, then maximum resection is probably advisable, especially if the patient is young and has a good performance status. Complications of surgery include disorders of eye movement, ataxia, and cognitive problems. If a supratentorial approach to the tumour is taken, there is a higher incidence of visual field defects and hemiparesis. Surgery for pineal region tumours is associated with a morbidity of 12 per cent and mortality rate up to 8 per cent (Bruce and Stein 1993). Pre-operative or 12 day post-operative spinal MRI may identify spinal seeding from the pineal region tumour. Standard radiation schedules for germinoma consist of 40 Gy to the whole brain and a boost to 55 Gy to the pineal region. Radiation therapy dosages of <50 Gy are associated with increased risk of recurrence (Schild et al. 1993). The need for craniospinal irradiation is uncertain, but most centres only suggest this for documented CSF metastases. Germinomas are also chemosensitive and chemotheraphy has been advocated in young patients in order to delay radiation or possibly reduce the dosage of radiation given. Germinomas and non-germinomatous germ cell tumours are sensitive to cisplatinum, plus or minus etoposide, and to cyclophosphamide, and pre-radiation chemotherapy is now being advocated by some centres (Patel et al. 1992). Pineoblastomas respond only partially to radiation and less well to chemotherapy. Various chemotherapy regimes have been tried with only limited success.

Craniopharyngiomas are benign tumours that usually present in childhood or early adulthood. They arise from the embryological remnants of Rathke’s pouch in the suprasellar area or pituitary region. Symptoms usually present in adolescence or early adulthood. In children the most common symptom is growth failure and adults present with sexual dysfunction in men and amenorrhoea in women. The tumour can present with hypopituitarism in 25 to 40 per cent, diabetes insipidus in 50 per cent, visual failure from pressure on the optic chiasm or optic nerves in 40 to 70 per cent, raised intracranial pressure from hydrocephalus from obstruction of the third ventricle in 20 to 40 per cent, or with personality and memory problems. The tumour is usually a mixture of cysts and solid components, where the cysts contain thick fluid like engine oil containing cholesterol crystals. Skull X-ray may demonstrate calcification in the suprasellar region, but MRI scan is the most valuable investigation (Fig. 27.2). The sagittal and coronal scans provide invaluable information to the surgeon. Differential diagnosis includes meningioma, optic nerve glioma, teratoma, dermoid or epidermoid cyst, metastasis or sarcoidosis.

After correction of any endocrinopathy, definitive operation can be performed in relative safety. The mainstay of treatment is resection of the tumour, although this has to be tempered by its tendency to be adherent to surrounding structures. Total resection is frequently impractical and attempts at aggressive resection have resulted in high morbidity and mortality of up to 20 per cent and recurrence rates of 30 to 40 per cent (Yasergil et al. 1990; Weiss et al. 1989; Wen et al. 1989). Most authors suggest safe subtotal resection and either post-operative radiotherapy to the residual disease or radiotherapy at the time of recurrence, depending on the age of the patient (Weiss et al. 1989; Wen et al. 1989). Some still suggest attempted complete removal as the best approach (Yasergil et al. 1990). There are no randomized studies of safe subtotal resection +/− radiation therapy versus maximum possible resection +/− radiation therapy and no randomized studies of early radiation versus delayed radiation. If radiation is given after resection, the usual advised dose is at least 54 Gy. At recurrence, it is not uncommon to get enlargement of one of the cysts. Cystic recurrences can be treated by placement of a reservoir and aspiration of the cyst intermittently (Gutin et al. 1980). A different approach is to instil ³²P a beta emitting isotope with limited penetrance. This has been reported to result in cyst regression in >80 per cent of cases with good symptomatic relief (Pollock et al. 1995). Twenty year survival for craniopharyngioma approaches 60 per cent, but recurrence is common and morbidity is significant (Regine et al. 1992). Craniopharyngiomas probably have a better prognosis when diagnosed in adults than when diagnosed in childhood.

Improved treatment of systemic malignancies has lead to an increase in the frequency of brain metastases, possibly because the brain may act as a ‘sanctuary site’ for cancer cells during systemic chemotherapy (Greig et al. 1990). The incidence of intracerebral metastases is approximately 14/10⁵/year; therefore each year in the United Kingdom one would expect approximately 8000 new cases. Metastases account for 45 per cent of all intracranial tumours and 60 per cent of intracerebral tumours. The frequency of brain metastases varies depending on the primary tumour but ranges from 12–35 per cent of all cancer patients (Posner et al. 1978; Galicich et al. 1996). Brain metastasis are generally a late manifestation of cancer and systemic metastases frequently co-exist, however, 36 per cent of patients do not have a past history of cancer at initial presentation. It has been estimated that only 19 per cent of patients do not have metastases at other sites at the time of presentation, but this was based on a cancer hospital population (Cairncross et al. 1980). The frequency of isolated brain metastases is likely to be higher in a general hospital population study. Lung cancer, cancer of unknown origin, breast cancer and melanoma account for 90 per cent of brain metastases (Grant et al. 1996). It is very unusual for patients with breast or gastrointestinal tract malignancies to present with brain metastases as the initial manifestation of cancer. If there is no history of malignancy at presentation with brain metastases the primary site is most commonly lung, in 55 per cent, or the primary tumour is not identified prior to death in 40 per cent (Grant et al. 1996).

Seizures are the presenting symptom in about 16 per cent and will eventually occur in up to 40 per cent of patients at some stage (Posner 1980). A randomized controlled trial has failed to show any advantage to prophylactic prescription of anticonvulsants in patients with brain metastases (Glantz et al. 1996). Anticonvulsants should not be given prophylactically in patients with metastases. Anticonvulsants are helpful in controlling frequency of seizures. Patients with cancer are at an increased risk of developing deep venous thrombosis, because of immobility and possibly as a result of hypercoagulation related to cancer (Dhami et al. 1993). This risk may be increased if hemiparesis results from a brain metastasis. Patients at risk of deep vein thrombosis with multiple risk factors such as hemiparesis, cancer, or operation should be managed with elastic stockings and subcutaneous heparin (Monreal et al. 1996). A literature review of patients with cancer who had either prophylactic therapy for deep vein thrombosis or pulmonary embolus showed that prophylactic anticoagulation or vena caval filter did not improve quality adjusted life expectancy. However, anticoagulant therapy provided a 9 per cent gain in quality adjusted life expectancy for patients with acute deep vein thrombosis and a 16 per cent gain for patients who survived a pulmonary embolus, and vena caval filter yielded 11 and 18 per cent gains respectively (Sarasin et al. 1993).

Steriods Symptomatic management of brain metastases with steroids for headache, focal neurological deficit, or cognitive problems is very effective in the short term in reducing the effect of brain oedema. The mechanism of action of steroids remains uncertain. Steroids do reduce the amount of oedema around the tumour and repair a leaky blood brain barrier, but the symptomatic relief within 6–24 hours after starting antedates any obvious change on CT scan or MRI. Approximately 70 per cent of patients improve with steroids. A reasonable starting dose would be 4 mg of dexamethasone intravenously four times a day, then changing to oral medication and altering the timing to give the last dose before 6 pm, because insomnia is a particularly common side effect of treatment. Steroids usually reach their maximal effect by 7 days and a gradual reduction in dose is advised then whether or not there has been an improvement in neurological deficit. Steroids almost certainly will provide a slight survival benefit, although this has never been proven in any randomized controlled trial and probably will never be. Their long-term use however is limited by the growth of the tumour, and the systemic side effects of long-term treatment.

Prognosis In patients with cerebral metastases will depend on the age; Karnofsky performance score 70; control at primary site and systemic metastatic spread. These factors have been used in recursive partitioning analysis to form classes with quite different survival (Gasper et al. 1997) (Fig. 27.15). Patients with single metastasis, Karnofsky performance score 70, aged <65 years, with controlled primary and no extracranial disease have a median survival of 13.5 months compared with only 6 months if the brain metastases are multiple and <2.3 months if the performance score is < 70.

Untreated patients usually die in a median of 4 weeks. There is evidence that prophalactic radiotherapy in small cell lung cancer prevents development of brain metastases and extends survival as well without producing cognitive problems if the fraction size is kept below 3 Gy. In situations where brain metastases are present and long-term survival and therefore long-term cognitive damage is likely to be reduced, radiotherapy is associated with median survivals of 3 to 6 months. Case selection undoubtedly plays a large part in this; nevertheless, palliative whole brain radiotherapy is the treatment of choice for most patients with brain metastases, since most are multiple, in non-resectable sites, and most patients have systemic spread elsewhere. The most effective dose of radiotherapy is widely debated. There does not appear to be a difference between 20 Gy over one week; 30 Gy in 2 weeks and 50 Gy over 4 weeks (Berk 1995).

Single brain metastasis refers to a single metastasis from a systemic tumour irrespective of the extent of spread to other organs. The term ‘solitary brain metastasis’ refers to the brain being the only site of systemic spread. Solitary brain metastasis is uncommon but has a better prognosis, if the brain disease can be controlled. Approximately 30 to 40 per cent of cerebral metastases are single (Delattre et al. 1988). Metastases from colon, kidney, and breast are more frequently single than metastases from lung or melanoma, but because lung cancer has a higher incidence than colon or kidney, lung cancer remains the most likely cause.

Surgery There is no evidence that operation for a single brain metastasis extends survival in patients with active cancer at other sites. Nevertheless, most patients with known cancer and a presumed single brain metastasis should be considered for resection because of the higher radiological diagnostic error rate for single brain lesions. In one study of patients with known systemic cancer and a CT brain scan suggestive of single metastasis, 11 per cent were found to have a different histological diagnosis, frequently non-malignant (Patchell et al. 1990). If patients are being considered for surgical resection of a single brain metastasis, it is usually advisable to perform an MR brain scan, to determine whether there are actually multiple micro-metastases not identified on CT scanning (Fig. 27.16). The diagnostic accuracy approaches 95 per cent. Surgical resection of a cerebral metastasis is feasible in selected patients. A surgically accessible lesion can be defined as one that is superficial, that is close to the brain surface or abutting a fissure or sulcus, and can be operated on with minimal parenchymal resection. This type of metastasis can frequently be resected even in eloquent areas of the brain. Surgery has the benefit of removing the lesion, reducing the need for long term steroids, potentially improving quality of life and providing a small survival gain in certain situations. Three randomized controlled studies have examined the place of resection of a single brain metastasis in patients with stable disease elsewhere. Two have demonstrated that resection improves survival to 40 weeks versus 15 weeks and to 10 months versus 6 months respectively (Patchell et al. 1990; Vecht et al. 1993). The third failed to demonstrate any difference (Mintz et al. 1996). A Cochrane Review and meta-analysis did not demonstrate a statistically significant difference in survival between the two treatments Hazard Ratio 0.74 (95 per cent CI 0.39 to 1.40, p=0.35). It did suggest a possible improvement in functional independent survival with surgery (Hart and Grant 2005). The duration of functional independence was better in the surgically treated group and there were fewer deaths from neurological disease. The post-operative mortality rate was 4 per cent in each group in one study where biopsy and radiation was compared with resection and radiation (Patchell et al. 1990) and 9 per cent in the second study in the group treated by resection and radiation versus 0 per cent in the group who received radiation therapy only (Vecht et al. 1993). Is there a need for whole brain radiotherapy after surgical resection? One small trial of post-surgical radiotherapy versus no post-surgical radiotherapy demonstrated a reduction in radiological recurrence with whole brain radiotherapy and fewer cases of neurological deterioration; however, there was no difference in functional independence or survival (Patchell et al. 1998). Recursive partitioning analysis class 1 patients may best be managed by delaying whole brain radiotherapy, since they are at the greater risk of delayed radiation related side effects.

Radiotherapy The majority of patients with single brain metastasis will not be suitable for surgery, due to the tumour being inaccessible, systemic disease, or other health-related factors, and radiation remains the accepted palliative treatment for most patients. There is no doubt that certain patients can have a symptomatic and imaging response to radiation. These patients usually are younger than 60 years, have a good Karnofsky performance score of >70, radiosensitive tumours and controlled primary tumour, and metastatic disease confined to the brain. Failure of radiation to have any clinical or imaging response is more commonly seen elderly patients and patients with a Karnofsky performance scale <70 (Deiner-West et al. 1989). The optimal dose fractionation schedule for treatment of brain metastases remains uncertain and varies widely from 20 Gy given over 1 week to 50 Gy over 4 weeks. Conventional whole brain radiation therapy is thought to increase median survival in patients with brain metastasis by 3 to 6 months (Cairncross et al. 1980); however, this is based on retrospective non-randomized matched controlled series and is probably optimistic.

Technical advances in radiotherapy have re-opened the debate about the value of surgery for single brain metastasis. Stereotactic radiotherapy using a linear accelerator with capability for three dimensional conformal external radiation and a non-invasive removable frame, which allows fractionated treatments, or stereotactic radiosurgery using multiple cobalt-60 sources and a fixed rigid surgically attached stereotactic frame, the ‘gamma knife’ for single session treatment may be as effective as surgical resection for single brain metastasis. Radiosurgery is high dose single fraction external irradiation of a stereotactically well defined target. For technical reasons the metastasis must be <3 cm in diameter and ideally should not border the ventricles, brainstem, or cranial nerves. Radiosurgery is an option for treatment in patients with single metastasis who are unfit for surgery or have a metastasis in a surgically inaccessible site. Highly selected series have demonstrated local control in 80 per cent of treated cases and an incidence of radiation necrosis of approximately 5 to 10 per cent (Flickinger et al. 1994; Alexander et al. 1995). Stereotactic radiosurgery using the conventional LINAC system or gamma knife has probably superseded the use of interstitial brachytherapy, which requires the placement of radioactive implants into the bed of the tumour after surgical resection or by stereotactic implantation and has a high incidence of radionecrosis. Because of the possibilities of treating single brain metastasis with surgical resection or stereotactic radiosurgery, there is now debate whether there is any need to treat the whole brain, if they are truly single. Randomized controlled trials to examine surgical resection versus radiosurgery alone are underway. A recent study of patients with between 1 and 3 non-resectable brain metastases treated with whole brain radiation therapy with or without stereotactic radiosurgery boost demonstrated no increase in the overall survival. Overall survival was 6.5 months with stereotactic radiosurgery boost compared to 5.7 months for radiotherapy alone. Post-hoc subgroup analyses of patients with a single metastasis demonstrated a better survival with the addition of stereotactic radiosurgery of 6.5 months versus 4.9 months. There was no significant difference in the cause of death (Andrews et al. 2004).

Management of young patients with multiple brain metastases is different from patients with single metastasis. Firstly, there is less chance of misdiagnosis on CT or MR scan (Fig. 27.16); secondly, there is less chance of having two or three metastases at surgically resectable sites, and thirdly, there is usually less opportunity for stereotactic radiotherapy, because the metastases are multiple. Conventional whole brain radiation without histological confirmation is almost always the management of choice. In patients with multiple brain metastases but stable systemic disease, there is no evidence that surgery and cranial radiation is superior to radiation alone. There are highly selected reported cases of good symptom control and extended survival from operation on two or three brain metastases at surgically accessible sites although such cases are few and far between. However, the effectiveness of surgery for multiple cranial metastases remains highly debatable and survival is probably poorer than patients with single brain metastasis (Hazuka et al. 1993). The series suggesting an advantage to surgery for multiple brain metastases were poorly matched, naturally highly selective, did not compare with patients treated with whole brain radiotherapy alone, and it is difficult to determine whether resections were complete or not (Bindal et al. 1993; Hazuka et al. 1993).

Systemic chemotherapy in a very selected patient group with potentially chemoresponsive tumours, such as breast, small cell lung or germ cell tumours, may be offered and may improve systemic disease, but usually the blood brain barrier will limit the efficacy in patients with brain metastases (Kristjansen et al. 1988; Boogerd et al.

1992). Attempts to overcome the effect of the blood brain barrier by using fat-soluble chemotherapeutic agents or by giving a bradykinin analogue such as RMP-7, to increase the permeability of the blood brain barrier and allow chemotherapy such as carboplatin to cross the blood brain barrier have been tried with very limited success.

Management of elderly patients with multiple brain metastases and active systemic disease is palliative with aims being symptomatic control and supportive care. There is no good evidence that radiation extends survival in elderly, disabled patients with active primary disease. Predicted survival of this group without treatment is in the region of 2 months

Malignant meningitis is defined as diffuse or widespread multifocal neoplastic involvement of the subarachnoid space. It can be due to spread from primary central nervous system tumours, metastatic spread from systemic malignancies, or due to haematological malignancies (Table. 27.7) (Grossman et al. 1991; Walker 1991; Recht 1991). The pathogenesis is probably multifactorial. Haematogenous spread via the choroid plexus is considered to be the most common route of spread especially for haematogenous malignancies, although rupture of cerebral metastases or spread along perivascular spaces of perforating vessels is very likely in cases related to primary central nervous system malignancies and a percentage of cases with intraparenchymal metastases. In one study 50 per cent of patients with malignant meningitis from solid systemic malignancies had previously had intraparenchymal metastases (Grant et al. 1994). A further but less common possibility is spread from deposits in the subdural space or associated with epidural spinal cord compression and spread along the nerve roots. The dura is thick and acts as a strong physical barrier to direct spread but about 5 per cent of patients with epidural spinal cord compression have co-existing malignant meningitis.

Primary central nervous system malignancies that abut the ventricles or lie close to the surface of the brain are most likely to spread to the CSF. CSF spread occurs in 30 to 50 per cent of cases with medulloblastoma, 10 to 20 per cent of cases with ependymoma and 1 to 5 per cent of glioblastoma. This CSF spread is most commonly asymptomatic but ‘dropped’ metastases especially from tumours in the posterior fossa, such as ependymoma and medulloblastoma, can result in a cauda equina syndrome or spinal cord compression. Imaging of the spinal canal is an important investigation to consider prior to planning radiation therapy or further management.

Malignant meningitis is seen in 4 to 8 per cent of autopsied cases dying with systemic cancer. Approximately 5 to 10 per cent of patients with breast or lung primaries will develop malignant meningitis and these are the two most common primary sites in most series. Nevertheless the occurrence of malignant meningitis is higher in rare malignancies such as melanoma, in 10 to 15 per cent, and systemic lymphoma, especially non-Hodgkins lymphoma at 30 per cent. In haematological malignancies such as acute lymphocytic leukaemia, 40 per cent of patients have malignant cells in the CSF and in acute myelocytic leukaemia 7 per cent of patients have CSF involvement (Walker 1991).

Pathology Macroscopically, there is opacification of the meninges usually at the base of the brain but also over the convexities or cauda equina region. The pathological changes are similar to infective meningitis. Diffuse or multifocal tumour infiltrates occur with reactive fibrosis and lymphocytosis. It is not uncommon for the reactive lymphocyctosis found in the CSF to cause confusion between systemic malignancy with reactive lymphocytosis, lymphoma, or even infective meningitis. These infiltrates are commonly at the ventral surface of the brain and in the cerebral and cerebellar sulci. Tumour can encase the basal meningeal vessels causing ischaemia and infarction of the perforating vessels or encase the perineurium of the cranial or spinal nerves causing ischaemia and then degeneration. Occasionally tumour will invade nerves.

Clinical features The pathophysiology of malignant meningitis can be predicted from the pathology. Hydrocephalus resulting from obliteration of the foramina of Magendi and Lushka occurs rarely but slowed egress of CSF via the arachnoid villi can produce a communicating hydrocephalus with raised intracranial pressure. Interference with the blood supply to the parenchyma causes infarction. Metabolic competition between tumour and nerves may be the reason for the gradual onset of cranial neuropathy or radiculopathy although direct invasion of the nerves and parenchyma undoubtedly also occurs. Malignant meningitis is usually a late complication of cancer and often presents at the same time as advancing disease at other sites. Patients will often have already had intraparenchymal disease. The diagnosis of malignant meningitis should be considered particularly in patients with neurological symptoms or signs affecting a combination of cranial nerve, spinal root, and cerebral cortex (Grossman and Moynihan 1991; Walker 1991; Recht 1991). It has been estimated that 50 per cent of patients have mild memory impairment at diagnosis and dementia occurs in 30 per cent. Headache is a presenting feature in 40 per cent and usually comes on gradually but becomes increasingly severe and intractable. The characteristics of the headache may be those of raised intracranial pressure or meningeal or vascular headache. Focal or generalized seizures occur in 5 to 10 per cent of cases. Neurological signs are frequently asymptomatic, for instance, absent reflexes, mild weakness, or subtle sensory signs. The single most common feature is a cranial neuropathy occurring in 80 per cent. The extraocular muscles are most commonly affected, in 75 per cent, followed by the facial nerve in 47 per cent or acoustic nerve in 40 per cent. The optic nerve is involved in about 38 per cent with papilloedema in 19 per cent, and reduced visual acuity in 17 per cent. Spinal root disease is the presenting symptom in about 25 per cent of cases and can be associated with back pain, limb pain, numbness, or weakness. Commonly there is a mixture of upper and lower limb spinal root symptoms to be present at the same time. On examination at the time of diagnosis 80 per cent have weakness in one or more roots and 60 per cent will have absent reflexes at some level. Occasionally cauda equina symptoms occur with sensory deficits from L1 downwards.

Investigations depend on the site of involvement and the differential diagnosis. The two most useful investigations are an imaging investigation at the clinically determined level, gadolinium enhanced MRI scan (Fig. 27.17) or myelography and CSF analysis. T1-weighted MRI with gadolinium enhancement will reveal an abnormality in 30 to 70 per cent of cases, depending on the case series reported. CSF analysis will reveal an elevated intracranial pressure of >160 mm CSF in 45 to 65 per cent of cases. CSF may also reveal an elevated protein of >0.5g/l in 81 to 89 per cent, low CSF glucose of <2.5mmol/l) in 31–41 per cent and ‘lymphocytosis’ of >5 cells/mm³ but negative cultures in 54 to 72 per cent (Wasserstrom et al. 1982). It is important to send at least 5 ml of fresh CSF quickly to the laboratory or the likelihood of obtaining a diagnostic sample may be impaired. The first CSF sample will be positive for malignant cells in about 54 per cent of cases. This is substantially higher in diffuse cellular malignant meningitis, in 75 per cent, than multifocal nodular meningeal disease, in 38 per cent. Repeated lumbar puncture will identify a further 30 per cent of cases missed on the first. If two lumbar punctures do not demonstrate malignant cells, the likelihood of a positive cytological diagnosis reduces dramatically, with only 1 per cent subsequently being positive (Wasserstrom et al. 1982). In the presence of repeated negative cytological specimens, the diagnosis can sometimes be made by cisternal puncture, in 2 per cent, or from sampling from ventricular CSF if there is a shunt or a ventricular access device, in 2 per cent. Approximately 10 per cent of cases will persistently have negative cytological CSF examinations, even in the presence of multi nodular deposits in the subarachnoid space. It is highly likely that the natural history, management plan, and response to treatment will be different in diffuse highly cellular malignant meningitis compared with the predominantly multi-nodular form. It may be possible to improve on the diagnostic accuracy and specificity of tumour type by using immunohistological staining methods on cytospun preparations of CSF. Epithelial membrane antigen, cytokeratins CAM 5.2, prostate specific antigen, and thyroglobulin can confirm the diagnosis and give clues to site of the primary if previously unknown. B and T cell markers may be supportive of lymphoma, HMB 45 is a relatively specific marker for melanoma and placental alkaline phosphatase may confirm germinoma. Glial fibrillary acidic protein positive staining of cells can demonstrate that the cells are of glial origin. Alphafetoprotein estimation and human chorionic gonadotrophin may support a diagnosis of teratoma or choriocarcinoma.

Management of malignant meningitis is controversial. In most cases CSF disease is a late pre-terminal manifestation of widespread disease. Treatment will depend on tumour type, with lymphoma and breast more responsive than melanoma and lung; extent of tumour spread systemically; whether the symptomatic site is cranial or spinal; raised intracranial pressure with hydrocephalus whether or not shunted; previous treatment; effectiveness of systemic chemotherapy; and meningeal deposits versus diffuse CSF pleocytosis. Most commonly, symptom control is all that can be reasonably offered using steroids, anticonvulsants and analgesia. If there is hydrocephalus, this can be shunted. It is uncertain how effective intra-reservoir chemotherapy treatment is in the presence of shunted hydrocephalus and the frequency of encephalopathy in this treatment group is higher. If there are symptomatic solid leptomeningeal metastases, radiation therapy occasionally stops progression or helps symptoms although relief is usually short-lived. If there are no solid leptomeningeal metastases, systemic disease is potentially treatable, and the patient is not severely impaired, then treatment of the CSF may be worthwhile. This is usually best given after placement of an Ommaya reservoir, since the distribution of chemotherapy is probably better than using the lumbar route (Shapiro et al. 1975). Nevertheless, placement of a reservoir and intra-reservoir treatment is not without complications. From a personal review of the literature, technical problems with placement occur in 6.5 to 28 per cent of cases, infection in 4.9 to 50 per cent, toxic complications of treatment in 1.7 to 20 per cent, and the mortality in reported series range from 0.5 to 8.3 per cent.

The only drugs routinely used are, methotrexate, cytosine arabinoside, or thio-tepa. In practice, if the systemic tumour is not sensitive to these agents then the CSF disease will not be sensitive. If CSF disease comes under control and malignant cells disappear from the CSF, the methotrexate can be reduced to once a week or discontinued, and clinical and CSF follow-up will determine whether further treatment is necessary. Intrathecal methotrexate should be given with preservative free saline and treatment with folinic acid, should be started at 15 mg orally every 12 hours on the day of treatment and for the following 24 hours. Folinic acid reverses the peripheral side effects and can prevent mucositis and marrow suppression. Methotrexate should be withheld if the white blood cell count falls below 3000/mm³ or platelets below 100 000/mm³. If leucoencephalopathy develops, methotrexate should probably be replaced by cytosine arabinoside 40 mg intraventricularly. There are approximately 5 per cent toxic deaths from treatment and 15 per cent develop confusion, disorientation, headache, nausea, or vomiting within 48 hours of methotrexate treatment although this usually resolves after 48 hours. Arachnoiditis and transverse myelitis can occur with methotrexate. The late complications of treatment with methotrexate include leucoencephalopathy and a necrotising encephalopathy. In certain situations, systemic chemotherapy may be valuable in treating the CSF disease or extending survival, for instance with high dose intravenous methotrexate for primary central nervous system lymphoma or breast carcinoma (Grant et al. 1994; Ackland et al. 1987). Liposomal cytosine arabinoside has been used in a controlled trail against soluble cytosine arabinoside and has been shown to have a better cytological response rate of 71 per cent versus 15 per cent (Glantz et al. 1999). Intrathecal topotecan has also been used recently (Blaney et al. 2003).

Prognosis For treated malignant meningitis this varies from series to series and one suspects that this is an area where ‘gearing’ of results and publication bias plays a large role in the apparent effectiveness of treatment. CSF becomes negative for malignant cells in approximately 40 per cent of cases. In general 25 per cent of patients have symptomatic improvement, 50 per cent remain stable for short periods of several weeks and 25 per cent progressively decline and die in 6 weeks. Periods of stability and improvement range from 1 week to 2.5 years with a median of 3 months. Median survival for untreated malignant meningitis is approximately 4 to 6 weeks. With ‘aggressive’ treatment one-third are dead in 6 weeks, median survival ranges from 9 to 24 weeks and 10 per cent survive >1 year. In selected cases where there is no systemic disease, two-thirds of patients will remain stable or improve, with median survival of 10 months, and 20 per cent are alive at 1 year (Kim and Glantz 2001).

The complications of surgery for brain tumours will depend on whether the tumour is sited intracerebrally, intracranial extracerebrally, or in the neck or skull. Complications will also depend on surrounding important structures such as nerves, endocrine or vascular structures, the radiological appearance of the tumour regarding size, uniformity, or necrotic areas, tumour pathology, the experience of the neurosurgeon, and the state of general health of the patient.

It is unlikely that the frequency of side effects recorded in the literature reflects the day to day frequency of side effects in general neurosurgery units (Maurice-Williams 1997). The complications of neurosurgery can be divided into non-surgical and surgical complications.

Patients with cancer are at an increased risk of deep venous thrombosis and pulmonary embolus. The additional risks of a surgical operation with bedrest and possible intra-tumoural bleeding make peri-operative management difficult. Deep venous thrombosis and pulmonary embolus pose a serious risk and prophylaxis with elasticated stockings and subcutaneous heparin are usually indicated (Frim et al. 1992). Electrolyte disturbances secondary to diabetes insipidus or syndrome of inappropriate antidiuretic hormone secretion can lead to a stormy peri-operative course. Patients may be systemically unwell as a result of malignancy or super-added chest or urinary tract infections. Endocrine deficiencies should be treated prior to neurosurgery on the pituitary gland and close attention paid to any peri-operative endocrine complications.

Post-operative haematomas at the operative site occur in approximately 5 per cent of patients. This frequency is probably higher in patients with malignant melanoma, choriocarcinoma, lung carcinoma, glioblastoma and lymphoma. This higher risk of bleeding may influence decision on whether stereotactic biopsy or an open procedure is performed in some cases. In operations for an intracerebral tumour, a transient neurological deficit will occur in approximately 10 per cent of patients post-operatively, slightly less with stereotactic operations. There is recovery of the neurological deficit in approximately 50 per cent of cases. There is also a risk of seizures as a result of operation in those who have no prior history of seizures and sometimes a flurry of seizures occur in the post-operative period. Intra-operative stroke will occur in less than 1 per cent depending on selection. Tumours in the region of the sylvian fissure are probably best operated on by an open procedure, because of the moderately high risk of damaging of the branches of the middle cerebral artery with catastrophic results. Post-operative infective meningitis and cerebral abscess are rare complications of craniotomy but still occur. Post-operative hydrocephalus is also uncommon but occurs particularly in patients undergoing posterior fossa surgery, where hydrocephalus is probably due to post-operative brain swelling or contamination of the CSF by blood or debris. Operations in the region of the temporal lobe can result in significant memory deficits which are sometimes not appreciated because pre-operative cognitive assessments are commonly not performed as post-operative bedside assessments more commonly concentrate on focal weakness or sensory impairments. Operations on the head and neck may damage cranial nerves, such as the facial nerve in parotid surgery, infra-orbital nerve in maxillary surgery, palatal and vocal cord paralysis due to damage to branches of the vagus in radical neck dissections or ipsilateral Horner’s syndrome.

Mortality from craniotomies for malignant glioma or metastasis approach 5 per cent, and for stereotactic neurosurgery are approximately 1 per cent, depending on the selection of patients and the experience of the surgeon (Cabantog and Bernstein 1994).

The toxic side effects of cranial radiation can be divided into local effects and central nervous system effects (Section 5.9). The side effects or radiation therapy depend on the dose fractionation schedule used, the natural history of the underlying disease and the likelihood of get a radiotherapeutic response.

Some people will feel nauseated about 30 minutes to 1 hour after treatments, and find that small meals with a low fat content are usually preferable to a large lunch or dinner. Patients will develop alopecia, but the degree and likelihood of recovery will depend on the dose and fractionation schedule used. Hair loss starts about two or three weeks into treatment and maximum regrowth has occurred by 6 months. Frequently the hair that returns in the irradiated area is fine and curly and may be of a slightly different colour. Skin can become dusky red, dry and itchy about three weeks into treatment and slight deafness can occur due to wax build up. Most people feel tired and sleepy at the end of a course of radiation and some feel sick.

Some years after cranial irradiation there may be further local neuroendocrine or neural complications. Pituitary failure can occur if the pituitary has received a moderately large dosage directly for pituitary adenoma, or is in the treatment field as in fronto- temporal low grade glioma. Radiation usually affects the prolactin and sex hormones first, with an increase in prolactin levels and decrease in FH and LSH, causing problems with periods or infertility, then the thyroid stimulating hormone falls producing secondary hypothyroidism. If the optic nerve is in the treatment field, one commonly finds an afferent pupillary defect with optic neuropathy which is usually asymptomatic or only produces mild visual acuity disturbance. Years after temporal lobe or posterior fossa irradiation one may find mild sensorineural hearing loss.

The most serious central nervous system complications of radiation to the nervous system are: acute encephalopathy; subacute or early delayed demyelination; delayed cerebral radiation necrosis; and chronic leucoencephalopathy.

Acute encephalopathy is rare, but comes on usually within 24 hours of cranial irradiation. Symptoms consist of headache, nausea vomiting, fever or worsening of neurological deficits. Occasionally, swelling causes cerebral herniation. The likelihood of developing the acute encephalopathy is related to dose and whether the patient is pre-treated with steroids prior to radiation. It can be difficult in some patients with brain tumours to know whether the deterioration is attributable to radiation or to progression of the underlying tumour. The treatment is steroids and consider reducing the fraction size of radiation.

Early delayed reaction is common and is seen 4 weeks to 3 months after completion of cranial radiation. In patients with cerebral tumours the symptoms are indistinguishable from those of tumour progression, except there is usually commonly a feeling of excessive tiredness and nausea. In cases who have died with early delayed subacute radiation reaction and who have had a post-mortem, there are changes in the white matter of the brain stem or cerebrum of demyelination, similar to those of multiple sclerosis. The treatment is to re-institute steroids for a period of 4 to 8 weeks and the gradually reduce them and discontinue.

Delayed cerebral radiation necrosis is infrequent and can start months or years after cerebral irradiation. In patients who have cerebral tumours, the clinical and radiological appearances mimic tumour recurrence. MRI cannot adequately distinguish active tumour from radiation necrosis. Positron emission tomography and single photon emission tomography can sometimes give an indication of whether there is increased radioisotope uptake consistent with active tumour or reduced uptake related to an avascular mass consistent with radiation necrosis, but neither technique is infallible and the only sure way to find out is to resect the necrotic mass. In patients with malignant glioma, it is common to see areas of necrosis within the tumour consistent with radiation damage and other areas of active tumour. Radiation necrosis is characterized by fibrinoid necrosis, luminal narrowing or occlusion, medial fibrosis and adventitial proliferation in small arteries. There may also be bizarre, multinucleated astrocytes and foci of necrosis. The necrosis is thought to be due to ischaemia secondary to changes in the small and medium vessels. Other hypotheses for the necrosis are that the radiation-induced changes in the glia produce demyelination and white matter damage or that the radiation causes release of cytokines into surrounding brain which results in tissue damage. Necrosis as a result of tumour progression does not have the same degree of small vessel occlusive and fibrotic changes but has significant endothelial proliferation.

Chronic leucoencephalopathy is usually only found in long-term survivors of cranial irradiation. Ten per cent of patients who survive for more than 1 year after radiotherapy for cerebral metastases will develop cognitive problems. Relatives notice that the patient may lack motivation, there is psychomotor retardation, memory impairment, and ataxia or apraxia of gait. As time passes there may be urinary incontinence, marked dementia, inability to walk due to apraxia or ataxia and cortical myoclonus. High dose and large fractionation schedules are thought to be associated with a higher incidence of radiation-induced leucoencephalopathy (DeAngelis et al. 1989; Klein et al. 2002). Recent studies however have suggested that radiotherapy in glioma is not the main reason for cognitive deficits (Taphoorn and Klein 2004). The tumour itself and other medical treatments also contribute. The CT or MR scan shows diffuse white matter changes in the cortical white matter and ventricular dilatation or generalized atrophy. The clinical picture is similar to normal pressure hydrocephalus or basal ganglia disease. Patients do not improve with lumbar puncture or ventricular shunting.

Side effects of chemotherapy may be a property of the mode of delivery or of the agent itself. Modes of delivery of chemotherapy include direct chemotherapy into the tumour bed at the time of surgery, intra-arterial chemotherapy where the tumour is confined to an area supplied by one artery, usually the internal carotid, or systemic chemotherapy either intravenously or orally. Intra-arterial chemotherapy requires arterial catheterization and commonly a general anaesthetic since there can be severe pain in the distribution supplied by the sensory fibres within the artery. This is a direct toxic effect on the artery. In addition even with supra-ophthalmic instillation of chemotherapy, there can be significant optic nerve toxicity due to turbulence and back flow along the ophthalmic artery. This can result in unilateral visual loss or ocular necrosis. Intravenous chemotherapy must be given cautiously in a fast flowing arm vein. If chemotherapy gets into the soft tissues of the arm it can result in a severe local thrombophlebitis.

The other effects of systemic chemotherapy relate to the toxic effect of the individual drugs. Nitrosoureas such as BCNU cause nausea that starts about two hours after starting an infusion that may persist for 24 to 48 hours. Facial flushing or dizziness may occur during infusion which is rate dependent and resolves on stopping the drug. Bone marrow suppression occurs with almost all agents and is maximal 4 to 6 weeks after receiving the drug and usually settles by 8 weeks. Risks of infection, bleeding and tiredness are greatest around 4 weeks after treatment. Lung toxicity usually starts after a total dose of 1 g of BCNU. A restrictive ventilatory defect is found and it is valuable to monitor vital capacity regularly in patients who receive >1 g total dose. Renal function should also be monitored. Procarbazine can also cause haematological and gastrointestinal symptoms, but in addition can cause flu-like symptoms, rash, and neurological symptoms of ataxia, headaches, paraesthesia, and dizziness. Procarbazine can cause hypertensive crisis and severe gastrointestinal symptoms if given with antidepressants, alcohol, or tyramine rich foods such as cheese or bananas. Vincristine causes neurotoxicity with neuropathy, myopathy, and autonomic disturbance, gastrointestinal symptoms, and sometimes alopecia. Haematological toxicity is mild.

Steroid use can result in weight gain, oedema, electrolyte problems, diabetes, osteoporosis, thinning of skin, and predisposition to infections and peptic ulcers. Cisplatinum derivatives cause neurotoxicity with peripheral neuropathy or deafness, renal toxicity, and bone marrow suppression. High dose cytosine arabinoside and 5 fluorouracil can cause reversible cerebellar ataxia, in addition to bone marrow suppression and gastrointestinal and liver toxicity. Alopecia and infertility can occur with any of the drugs. Methotrexate can cause bone marrow suppression, mucositis, and rarely pneumonitis. It is nephrotoxic and hepatotoxic and can cause an encephalopathy if given in high dose. Toxic effects to intrathecal methotrexate are mentioned in Section 27.10.

Recovery and rehabilitation from a diagnosis of brain tumour or from medical treatment of a brain tumour may involve many specialists and support services. The first step is often coming to terms with the diagnosis and this can be eased by accurate, understandable medical information about the disease and its treatment options. This is best done by a doctor experienced in managing patients with brain tumours (Richards 1990). The Royal College of Physicians have produced guidelines regarding breaking bad news to the patients with malignant glioma and their relatives. (Davis et al. 1997). Written information can often be a helpful reminder to patients and relatives, particularly where there are cognitive difficulties. There are also helpful general informational leaflets about different sorts of brain tumours from charitable organizations. Occasionally, patients with brain tumours will have disabling anxiety or will become clinically depressed and may require special counselling or antidepressant medication. Neuropsychological symptoms may be related to the disease, for instance seizures or fronto-limbic involvement with tumour, treatment with anticonvulsants, steroids, radiation therapy or chemotherapy, or fear of the future including loss of health, independence, work, family position, or relationships. Neurocognitive support from psychologists, psychiatrists, support groups, and self help groups may all play a role in neuropsychological recovery by providing information, enhancing personal control and teaching coping mechanisms.

Physical rehabilitation involves improvement of neurological impairment by steroids, anticonvulsants, painkillers, or speech therapy, coping with physical disability using walking aids, eating aids, adaptation of the home and by maximizing physical independence thus reducing handicap by encouraging re-integration into home, work, and past-times where feasible, and minimizing unnecessary hospital contact. In the early post-operative period, physical rehabilitation usually progresses alongside medical therapy in hospital, however, there may be a feeling of active treatment for the tumour grinding to a halt after radiation and this is paradoxically sometimes a period of despair and anxiety while patients await ‘what’s next’. Patients should be made aware of possible early delayed effects, to allay the fear of early recurrence and should have a target directed plan for recovery which includes their own programme for rehabilitation and for periods of rest.