Cancer biology 
Cancer stem cells 
Cancer angiogenesis 
Tumor invasion & metastasis 
Principles of cancer screening 
Tumor markers 
Circulating tumor cells 
Breast cancer 
CNS tumors 
Lung cancer 
Gastric cancer 
Urologic tumors 
Liver tumors 
Pancreatic cancer 
Hereditary cancer syndromes 
Cancer complications 
Web resources 






The different histologic types of CNS tumors are shown in Table 1. Meningiomas, glioblastomas, and astrocytomas constitute more than half of all CNS tumors. The frequency of different histologic tumor types varies with age, as shown in Figure 1. The incidence of all brain tumors is highest in the 75- to 84-year-old group. The incidence of meningiomas increases with increasing age, whereas for gliomas and pituitary adenomas, the incidence increases with age but then declines at the highest age category (see Fig. 1A). Certain histologic types, such as germ cell tumors, medulloblastomas, and pilocytic asctrocytomas, are far more common in children than in adults (Table 2; see Fig. 1B).

Table 1. Primary Central Nervous System Tumors: Cell of Origin and Distribution by Histologic Type


Histologic Tumor Type



Cell of Origin

In All Age Groups[*]

Among Young Adults[†]


Arachnoidal fibroblast







Other astrocytomas












Embryonal tumors, including medulloblastoma




Pituitary adenoma





Cells from Rathke's pouch



Nerve sheath

Schwann cell







All others




 Choroid plexus papilloma or carcinoma

Choroid epithelial cell




Endothelial cell



 Germ cell tumor

Primitive germ cell




Pineal parenchymal cell




Notochordal remnant



* Frequency among all patients with primary brain and CNS tumors (N = 63,698). Gliomas (glioblastomas, astrocytomas, oligodendrogliomas) and ependymomas and neuroepithelial tumors account for 40% of all tumors and 78% of malignant tumors.

Frequency among young adults (20 to 34 years of age) with primary brain and CNS tumors (N = 5741). Data from Central Brain Tumor Registry of the United States (CBTRUS), 2005–2006.

Figure 1.  Age-specific incidence of primary CNS tumors by histologic type. A, Selected histologic types among all age groups. B, Selected histologic tumor types in children.  

Table 2. Primary Central Nervous System Tumors of Childhood: Distribution by Histologic Type


Histologic Tumor Type


Age 0–14 Years (N = 4214)

Age 15–19 Years (N = 1241)

Pilocytic astrocytoma






All other astrocytomas









Embryonal tumors, including medulloblastoma



Pituitary adenoma






Germ cell tumor



All others



Gliomas account for 56% of all tumors and 67% of all malignant tumors in children 0 to 14 years of age and for 45% of all tumors and 68% of all malignant tumors in those 15 to 19 years of age

For most patients with primary CNS tumors, the initial therapy usually is operative. Radiation therapy often is an important component of treatment after surgery, but for management of malignant disease, chemotherapy has had an expanding role. A great deal of knowledge has been amassed over the past 2 decades regarding the biology of brain tumors, which with time should lead to improved treatments.


An estimated 43,800 new cases of primary CNS system tumors were diagnosed in the United States in 2005. Approximately 18,500 were malignant, representing 1.35% of all cancers diagnosed that year. Malignant CNS tumors caused approximately 13,000 deaths in 2005.

On the basis of data provided by the Surveillance, Epidemiology, and End Results (SEER) Program, Deorah and colleagues found that the incidence of brain cancer increased until 1987, when the annual percentage of change reversed direction. The elderly experienced an increase in brain cancer until 1985, but their rates were stable thereafter. Overall, however, the incidence of glioblastoma has been increasing, with survival unchanged over the past 2 decades.

Ionizing radiation is one of the few factors shown to have a strong association with the development of brain tumors. Exposure to ionizing radiation represents the most important exogenous risk factor for childhood brain tumors. Prenatal diagnostic x-ray exposure increases the risk of childhood brain tumors, and various reports describe the occurrence of gliomas, meningiomas, and other brain tumors in children who received radiation therapy to the head for tinea capitis and for prior malignancies. A dose of 1 to 2 Gy of radiation, used to treat tinea capitis in Israeli children, was associated with an increased risk of developing brain tumors—specifically, meningiomas, gliomas, and nerve sheath tumors. A dose-response correlation in the induction of brain tumors was seen with a relative risk of 3.0 at a dose of 1 Gy. Tumors developed at least 6 years after irradiation, with a mean interval greater than 15 years. Even lower doses of radiation delivered with 226Ra used to treat hemangiomas in Swedish infants (mean dose to brain of 7 cGy) were found to be associated with an excess risk of intracranial tumors, including pituitary adenomas, gliomas, meningiomas, and nerve sheath tumors.

A large amount of data has been accumulated on the incidence of brain tumors in patients who received cranial irradiation for the treatment of acute lymphoblastic leukemia (ALL). The estimated cumula tive risk of secondary malignant brain tumors after childhood ALL therapy is 0.5% at 10 years after completion of therapy. In a study from St. Jude's Children's Hospital, the actuarial 20-year probability of developing a brain tumor in these patients was 1.4%. The probability of developing a high-grade glioma was greater in children younger than 5 years of age at diagnosis than in those 6 years of age or older (1.08% versus 0.45%; P = 0.045). An apparent dose-response correlation was observed: The 20-year risk of developing a brain tumor was 3.2% in patients who received greater than 30 Gy, versus 1.03% in those who received 21 Gy or less (P = 0.015). No CNS malignancies were seen in patients who did not receive cranial irradiation. The latency between irradiation and the diagnosis of a brain tumor ranged from 5.9 to 29 years (median, 12.6) but was longer for meningiomas (median, 19 years) than for high-grade gliomas (median, 9.1 years). Very similar results regarding the frequency of brain tumors, latency, and dependence on prior cranial irradiation were seen in studies from the German BFM (Berlin-Frankfurt-Munster) group and the Children's Cancer Study Group (CCSG). The types of brain tumors that have been reported in these series include gliomas, meningiomas, and medulloblastomas. Patients who received cranial irradiation for ALL also often received intrathecal chemotherapy. It has been suggested that cranial irradiation and intrathecal chemotherapy may work synergistically to increase the incidence of glial tumors.

Viruses can induce brain tumors in animals in the experimental setting; however, no conclusive data point to viruses as a cause of brain tumors in humans. Although many chemicals can induce brain tumors in laboratory animals, no definitive associations have been found in humans. For example, N-nitroso compounds, which commonly are present in foods, are known to be neurocarcinogenic in animals. Oxidants in the environment can cause DNA damage, so it has been hypothesized that antioxidants, such as vitamin E, found in certain foods may protect against the development of cancers. Epidemiologic studies, however, have provided mixed support for the idea that the intake of N-nitroso compounds, antioxidants, or specific nutrients in foods can influence the risk of developing brain tumors.

Also lacking is conclusive evidence that occupational exposure to industrial chemicals leads to the development of brain tumors, although a number of studies have suggested such a link. Some of the chemicals that can induce brain tumors in laboratory animals, such as polycyclic aromatic hydrocarbons, can do so only when administered by direct contact or transplacentally, but not by inhalation or dermal contact; the latter two modes of exposure are more relevant in the occupational setting. Specific chemicals that have been examined include cosmetics containing N-nitroso compounds, organic solvents, chemicals used in the manufacture of synthetic rubber, formaldehyde, phenols, polycyclic aromatic compounds, polyvinyl chloride, and pesticides. Vinyl chloride can induce brain tumors in rats; however, a recent review found that the association in humans is inconclusive. Bohnen and Kurland reviewed studies examining the incidence of brain tumors in agricultural workers exposed to pesticides and found that the results were inconclusive. A large meta-analysis of studies examining workers in the petrochemical industry found no increased risk of brain tumors in this population.

Recently a great deal of interest has emerged in a possible association between use of cellular telephones and the risk of brain tumors. In a case-control study, Inskip and coworkers were unable to show a correlation between the duration of cell phone use and the development of gliomas, meningiomas, and acoustic neuromas. Other large case-control studies also have failed to find any association between cell phone use and the risk of developing brain tumors. Nevertheless, some still claim that there is a link between cell phone use and brain tumors.

Other factors that have been analyzed for their possible relationship to the development of brain tumors include a history of head trauma and injury, drugs and medications, allergies, seizures, smoking and alcohol consumption, and exposure to power-frequency electromagnetic fields. None of these factors, however, have been shown to be conclusively important.

Most brain tumors represent sporadic cases; however, familial clustering has been noted. It is estimated that hereditary syndromes account for 2% of childhood brain tumors, although this may be an underestimate because hereditary syndromes may go undiagnosed in a number of cases. Some hereditary syndromes known to be associated with brain tumors are listed in Table 3. Some of the associations are extremely strong. Nearly 70% of all optic pathway gliomas occur in patients with neurofibromatosis type 1 (NF-1), and acoustic neuromas commonly occur in patients with NF-2. In addition, hereditary immunosuppression disorders such as Wiskott-Aldrich syndrome, as well as treatment-associated immunosuppression as in organ-transplant recipients, or exogenous immunosuppression as in human immunodeficiency virus (HIV) infection, are known to be associated with an increased risk of primary CNS lymphoma.

Table 3. Hereditary Syndromes Associated with Brain Tumors


Associated CNS Tumors


Chromosomal Locus

Defective Protein and Normal Function

Neurofibromatosis type 1

Optic pathway gliomas, meningiomas, neuromas



Neurofibromin; GTPase-activating protein (GAP) that negatively regulates Ras

Neurofibromatosis type 2

Bilateral acoustic neuromas, meningiomas, gliomas



Merlin; related to membrane cytoskeleton linker protein 4.1 superfamily

Tuberous sclerosis

Cerebral hamartomas





Subependymal giant cell astrocytoma (SEGA)



Tuberin; associates with hamartin; both are involved in signaling downstream of Akt

von Hippel-Lindau syndrome




VHL protein; degrades HIF1a

Li-Fraumeni syndrome

Malignant gliomas



p53; maintains genomic stability

Cowden's syndrome




PTEN; lipid phosphatase, counters PI3 kinase activation

Gorlin's syndrome (nevoid basal cell carcinoma syndrome)




Cell surface receptor; regulates normal brain development

Turcot's syndrome




APC; part of β-catenin/Wnt signaling pathway


Malignant gliomas



Involved in mismatch repair


Malignant gliomas



Involved in mismatch repair

Familial retinoblastoma




Rb protein; regulates entry into S phase


CNS lymphoma



ATM protein; involved in DNA damage sensing

Multiple endocrine neoplasia syndrome 1 (MEN-1)

Pituitary adenomas




APC - adenomatous polyposiscoli; ATM - ataxia-telangiectasia mutated; CNS - central nervous system


In the past decade, the field of oncology has seen explosive growth in elucidation of basic biologic processes. Normal cells have constraints in their ability to grow; for example, their growth generally is inhibited when they contact other cells. Tumor cells, by contrast, have sustained genetic mutations that allow them to overcome these constraints. Some of these changes allow them to proliferate in the absence of external cues. Other genetic changes allow them to invade adjacent normal tissues. Still other mutations allow tumor cells to induce angiogenesis and develop their own blood supply.

Cell Proliferation

Normal cells rely on growth factors secreted in their local environment to stimulate their growth. Many CNS tumors, however, have developed the ability to express their own growth factors along with the respective receptors, resulting in an autocrine loop that allows for self-stimulation. Platelet-derived growth factor receptor-α (PDGFR-α), for example, is overexpressed in all grades of astrocytomas, but  only higher-grade tumors overexpress the ligands PDGF-A and PDGF-B. Insulin-like growth factors and their receptors both are expressed in brain tumors including gliomas and meningiomas. The epidermal growth factor receptor (EGFR) is amplified or overexpressed in 50% of glioblastomas, and expression of transforming growth factor-α (TGF-α), a ligand that binds to this receptor, is increased in many gliomas. Both scatter factor (also known as hepatocyte growth factor [HGF]) and its receptor c-Met are expressed in gliomas, with the highest level of expression seen in the most malignant tumors.

As a result of increasing expression of receptors and ligands, increased signaling occurs in many brain tumors, resulting in activation of many different pathways that are important in proliferation. The best-studied of these pathways is the microtubule-associated protein (MAP) kinase pathway, which involves Ras and Raf. Another pathway that has attracted much attention recently is the phosphoinositide-3 (PI3) kinase pathway, which leads to activation of Akt. Mutation of PTEN, which occurs in 30% to 40% of glioblastomas, also can lead to increased Akt activation in these tumors.

Ras activation commonly is seen in human astrocytomas and neurofibromas despite the fact that these tumors rarely contain Ras mutations. In astrocytomas, Ras activation probably occurs through activation of growth factor receptors such as EGFR and PDGFR. Other mechanisms of activation of aberrant G proteins have been identified in other CNS tumors. In NF-1, loss of expression of neurofibromin, an inactivator of Ras, is seen (see Table 3). This loss of neurofibromin leads to the increased Ras activation seen in NF1-associated astrocytomas. Pituitary adenomas often show activation of the a subunit of the large heterotrimeric Gs protein, resulting in mitogenic signaling.

Cell proliferation is intimately tied to cell cycle regulation. For many cancers, mutations in two different pathways are important for deregulating the cell cycle. The first of these is the p16/Cdk4 or Cdk6/cyclin D/Rb pathway. The second is the p21/p53/Mdm/p19ARF pathway. Mutations in both of these pathways are common in many brain tumors, particularly gliomas.


Many brain tumors, particularly gliomas, display an invasive phenotype, with infiltration of tumor cells into surrounding tissues, making a cure very difficult to achieve. In fact, tumor recurrence was reported in one case even after the drastic measure of taking out the entire hemisphere in which a glioma was located. Numerous molecules associated with invasion have been found to be upregulated in gliomas, including tenascin-C, secreted protein acidic and rich in cysteine (SPARC), various integrins, and matrix metalloproteinases.

Angiogenesis and Hypoxia

For a tumor to grow beyond a certain size, it must develop a blood supply. The process of angiogenesis is described in detail in Chapter 8 . A number of growth factors are known to be important in angiogenesis. The most prominent of these is vascular endothelial growth factor (VEGF), which is overexpressed in many brain tumors. In one study, increasing VEGF expression correlated with increasing malignant grade in astrocytomas, oligodendrogliomas, and ependymomas. This study also found that increased expression of the VEGF receptors Flt-1 and KDR in tumor vasculature was found to correlate with increasing VEGF expression and malignant grade. Growth factors other than VEGF that may play a role in angiogenesis in gliomas include members of the TGF-β family, PDGF, placenta growth factor, basic fibroblast growth factor, and scatter factor/hepatocyte scatter factor (HSF). Angiogenesis can also be negatively regulated by factors such as thrombospondins 1 and 2 (TSP1 and TSP2). TSP1 is positively regulated by p53, so loss of p53, which commonly occurs in gliomas, can lead to decreased TSP1 expression and increased angiogenesis.

Angiogenesis is particularly prominent within glioblastomas. One of the characteristic features of these tumors is endothelial proliferation and neovascularization. A variety of growth factors that can increase angiogenesis are expressed by glioblastomas, foremost being VEGF. VEGF, also known as vascular permeability factor (VPF), is a potent inducer of capillary permeability. The high levels of VEGF expression in glioblastoma may be responsible for the edema associated with these tumors. Some evidence indicates that genetic changes common to glioblastomas, such as EGFR activation and PTEN mutation, may contribute to high levels of VEGF expression, perhaps through activation of the PI3 kinase pathway.

Despite expressing high levels of VEGF, glioblastomas contain significant regions of hypoxia, which may be a cause of treatment resistance. Hypoxia has been shown to be present in malignant gliomas by both polarographic needle electrode measurement and binding of the 2-nitroimidazole EF5. Hypoxia as measured by needle electrodes was not prognostic for survival. The latter study, however, found a correlation between more rapid tumor recurrence and hypoxia as measured by EF5 binding.

The presence of hypoxia in glioblastomas is consistent with the histologic features commonly seen in these tumors, including the presence of pseudopalisading necrosis and proliferative blood vessels. It has been proposed that pseudopalisades seen within glioblastomas represent a wave of tumor cells actively migrating away from central hypoxia that arises secondary to vaso-occlusion and intravascular thrombosis. It may seem counterintuitive that hypoxia can persist in the presence of high VEGF levels and robust neovascularization; however, the explanation may be that although hypoxia may stimulate VEGF and the formation of new vessels, many of these vessels are nonfunctional and do not transport oxygen well.

Stem Cells

The stem cell hypothesis proposes that a small fraction of cells with stem cell properties are the source of renewal of the tumor and determine a tumor's behavior. Such stem cells have been identified in human brain tumors and express the neural precursor marker CD133. A number of strategies may be used to target these stem cells. It recently has been demonstrated that these stem cells may be relatively radioresistant as a result of overexpression of DNA damage repair proteins including the checkpoint kinases Chk1 and Chk2. The cells may be made more radiosensitive by using specific inhibitors of Chk1/Chk2. Piccirillo and colleagues showed that CD133+ glioblastoma stem cells have a functional bone morphogenic protein (BMP) receptor pathway. By exposing mice implanted with orthotopic glioblastomas to BMP4, they were able to cause the CD133+ cells to differentiate toward glial cells and reduce tumor growth. These two strategies potentially may be used to target stem cells in glioblastomas. Another potential strategy is to target the tumor stroma, which provides a specialized niche for tumor cells. Calabrese and associates showed recently that stem cells in brain tumors occupied a perivascular niche and can be targeted by using antiantiangiogenic therapy.


Pathophysiology of Signs and Symptoms

Parenchymal brain tumors produce clinical signs and symptoms by three main mechanisms, each of which has important implications for therapy: (1) Infiltration along nerve fiber tracts is typical of low-grade astrocytomas and oligodendrogliomas. The first clinical manifestation may be a single seizure. Normal brain tissue may be present within areas appearing abnormal on MRI, and functional mapping may be necessary for safe resection of these slowly growing tumors.  (2) Displacement of brain tissue with production of vasogenic edema is typical of cerebral metastases. Such tumors sometimes can be resected or irradiated focally with less risk to adjacent normal brain tissue than is the case with infiltrating tumors. (3) Rapidly growing aggressive tumors such as high-grade astrocytomas may enlarge as a mass and also destroy surrounding neuropil to such an extent that surgical resection, although helpful for the reduction of mass effect and ICP, may not alleviate local symptoms and signs.

Intracranial neoplasms tend to produce progressive symptoms. Location and rate of growth determine both general and specific localizing symptoms and signs. Thus, a patient with a low-grade glial tumor may have seizures for many years or may exhibit behavioral alteration for many months before developing focal signs. With a more aggressive glial neoplasm, headache and focal signs may develop over a few weeks. Acute apoplectic onset is associated with hemorrhage or the development of hydrocephalus.

Brain tumor symptoms may be general, localizing, or falsely localizing. General symptoms include headache, lethargy, nausea, vomiting, and vague balance difficulties. These symptoms tend to be manifestations of increased ICP from a combination of expanding tumor volume and the production of associated vasogenic cerebral edema. Tumors also may cause raised ICP by obstruction of the ventricular system or blockage of the venous sinuses. Abrupt headache and exacerbation of neurologic signs may accompany the plateau waves of sudden increased ICP.

Sustained ICP in excess of 200 mm H2O causes brain shifts that can displace brain tissue through fixed intracranial openings, producing life-threatening herniation syndromes (Fig. 2; Table 4). The uncal herniation syndrome is caused by tumors arising in the lateral aspect of the brain, most commonly the temporal lobe. The earliest, most consistent sign of uncal herniation is a unilaterally dilated pupil due to compression of the ipsilateral third cranial nerve. This is followed by extraocular movement abnormalities consistent with an oculomotor palsy. The posterior cerebral artery may be compressed against the tentorium, leading to homonymous hemianopia. Brainstem compression can cause contralateral hemiapresis or, on occasion, ipsilateral hemiparesis. This is a result of compression of the contralateral cerebral peduncle against the edge of the tentorium, causing what is known as Kernohan's notch phenomenon. Patients with uncal herniation initially may be awake, but progression to obtundation, coma, and death may occur rapidly, within hours.

Figure 2.  Intracranial herniation syndromes evoked by supratentorial masses. The tumor and its edema (arrows) have produced the following (curved arrows): cingulate gyrus herniation under the falx cerebri; diencephalic herniation across the midline compressing the ipsilateral ventricle and producing the hydrocephalus in the contralateral ventricle; hippocampal gyrus herniation through the tentorial notch compressing the posterior cerebral artery and brainstem, and herniation of the cerebellar tonsils through the foramen magnum  

Table 4. Important Clinical Syndromes in Patients with Brain Tumors

Syndrome/Signs and Symptoms

Localization and Pathogenesis

Common Tumor Type(s)

Obstructive hydrocephalus



 Headache, papilledema

Posterior fossa

Medulloblastoma, ependymoma, astrocytomas

 Nausea, vomiting

Third ventricle compressed with growth of tumor or cyst


 Ataxia, stiff neck



Communicating hydrocephalus



 Headache/pressure waves

Arachnoid granulation

Meningeal gliomatosis, any primary or metastatic tumor type

 Gait apraxia

Scarring from treatment, hemorrhage, or infection


Venous sinus thrombosis



Uncal herniation



 Pupil dilation, oculomotor palsy

Temporal lobe herniates through tentorial notch, leading to compression of posterior cerebral artery, contralateral peduncle, midbrain

Any tumor type




 Ipsilateral hemiparesis






Central herniation



 Lethargy, small pupils

Diencephalic compression

Any tumor type

 Cheyne-Stokes respiration



Tonsillar herniation



 Posterior headache

Cerebellar tonsil herniates into foramen magnum

Posterior fossa tumor (astrocytoma, ependymoma, medulloblastoma)

 Stiff neck






Pituitary apoplexy



 Headache, diplopia

Hemorrhage into pituitary

Pituitary adenoma, meningioma, craniopharyngioma

 Third nerve palsy







The central herniation syndrome results from tumors that arise along the midline axis of the brain, especially those deep in the basal ganglia and thalamus regions. The initial signs and symptoms are due to diencephalic compression. The first evidence for this syndrome often is an alteration in the level of alertness and behavior. Some patients become agitated; others become very drowsy. Hemisensory or hemiparetic deficits and periodic Cheyne-Stokes respirations may develop in these patients. Initially with diencephalic compression, pupils are small (1 to 3 mm), but as the syndrome progresses and the midbrain and upper pons are compressed, the pupils dilate moderately and fix at midposition (3 to 5 mm). As the syndrome progresses, patients become progressively lethargic and apathetic and may develop Cushing's signs of hypertension and bradycardia due to direct compression of the hemodynamic control nuclei within the brain stem.

Tonsillar herniation may be caused by an expanding mass in the posterior fossa, the region of the brain between the tentorium and the foramen magnum. This results in the cerebellar tonsils being pushed through the foramen magnum. The syndrome is characterized by posterior headache, vomiting, stiff neck, and sometimes opisthotonic posturing. Other features may include dysconjugate eye movements and syncope with cough or sudden postural change. Frequently, these patients complain of visual dysfunction due to progressive papilledema affecting visual acuity. As a result of direct compression of the medulla and its respiratory center, irregular breathing and acute apnea may develop in these aptients. Further compression of the brainstem can lead to Cushing's signs of hypertension and bradycardia. Tumors in the posterior fossa may also completely obstruct the fourth ventricle, leading to an obstructive hydrocephalus, which, if left untreated, will manifest as a central herniation syndrome. From a clinical standpoint, tonsillar herniation can be very difficult to diagnose. Patients often are agitated at the onset of the syndrome and frequently are sedated with narcotics, which only further compromises respiratory function, with the potential for a fatal outcome.

These herniation syndromes can rapidly progress from the onset of symptoms to death. They can be precipitated by medical procedures. Lumbar puncture may lead to tonsillar herniation, and ventriculostomy may result in upward herniation in which the brainstem is forced up through the tentorial notch. A high index of suspicion is required in order to successfully diagnose and treat this condition by emergency intubation and administration of appropriate therapy (see later under Treatment of Brain Tumor Symptoms).

General Signs and Symptoms

Headache results from traction on pain-sensitive structures of the intracranial contents including the large cerebral vessels, the dura and meninges, the venous sinuses, and cranial nerves V and IX. Headache is the most common symptom of a brain tumor and occurs in approximately 50% of patients at some time during the course of the illness. Headache more frequently accompanies rapidly growing than slowly growing tumors. Tumors located in neurologically noneloquent brain areas such as the nondominant frontal and temporal lobes may manifest with headache as the sole clinical manifestation. The “classic” brain tumor-associated headache often is worse in the morning and lateralized to the affected side of the brain. Brain tumor-associated headache may be worsened by coughing or straining. A majority of patients with brain tumors, however, do not have these classic symptoms but instead have headaches that are deep, aching, and difficult to distinguish from tension-type headache. As a general rule, headaches from posterior fossa tumors are localized to the back of the head or neck, whereas headaches from tumors of the anterior and middle cranial fossas may be referred to the forehead or eye, at times being misconstrued as “sinus” headache.

Brain tumor-associated cognitive changes initially may be subtle and frequently are misdiagnosed as depression or, in the older patient, age-associated forgetfulness or early Alzheimer disease. The patient may complain of additional nonspecific signs and symptoms such as fatigue, concentration problems, irritability, and loss of interest in usual activities. Frontal lobe tumor location is the most common site for tumors producing these symptoms as the initial manifestation of neoplasm. Hypomania and psychosis are less common cognitive symptoms and when occurring as the presenting symptoms usually are associated with temporal lobe tumors.

Roughly paralleling cognitive decline and of poor localizing value by themselves, several other symptoms may emerge. Dysphagia is a common complaint, and caregivers may note that the cognitively impaired patient seems to take a long time to chew and swallow. Oral candidiasis is a potential complicating issue in patients on long-term corticosteroid therapy and appropriate treatment should be instituted. Similarly, incontinence tends to parallel the degree of cognitive impairment. Urinary retention may occur in patients with bifrontal brain disease or with spinal cord problems. Opiate medications may exacerbate the urologic problems.

Seizures occurring for the first time in adults are more likely to be due to focal brain pathology, particularly neoplasms, than seizures occurring in childhood. Intracranial tumors produce both generalized tonic-clonic seizures, secondarily generalized tonic-clonic seizures, and partial, localization-related seizures. The tumors that are most likely to manifest with seizures are slowly growing astrocytomas and oligodendrogliomas or oligoastrocytomas. Seizures occur at some time in 25% to 50% of patients with brain tumors. Tumors in the cortical and subcortical cerebral hemispheres, particularly the insula, are the most likely to produce seizures. Such seizures may be more difficult to control than those of idiopathic epilepsy and many patients require more than one antiepileptic drug (AED; see later under Treatment of Brain Tumor Symptoms). Seizure frequency may increase during radiation treatment and continue at an increased frequency for several months thereafter because of localized swelling.

Papilledema, swelling of the optic nerve head with engorgement of retinal veins, usually indicates raised ICP. In the pre-CT and MRI era, papilledema was a more frequent finding in brain tumors than it is today, because diagnosis now occurs earlier in the disease course. Currently, papilledema is seen in less than 20% of patients at presentation, down from 59% in the report by Huber.

The development of papilledema is dependent on the location of the tumor and the rate of growth. Papilledema is more common in patients with infratentorial tumors than in those with supratentorial tumors. Other papilledema-producing tumor areas include the third ventricle, cerebral aqueduct, and fourth ventricle. Thus, in adults, medulloblastomas, glial tumors of the cerebellum, hemangioblasto mas, and tumors of the cerebellopontine angle are most commonly associated with papilledema.

Vomiting, with or without nausea, may occur as a result of direct simulation of emetic centers in the floor of the fourth ventricle. This mechanism explains the nausea and vomiting seen with raised ICP, particularly when the rise in pressure has been rapid and is associated with hemorrhage or herniation. Patients with posterior fossa tumors frequently experience nausea and vomiting. More commonly, however, nausea is a nonlocalizing symptom, so the differential diagnosis must include adverse drug reactions from AEDs, analgesics, or other concurrent medications and gastritis from high-dose corticosteroid treatment.

Localizing Signs of Intracranial Tumor

Focal clinical signs of intracranial tumor reflect the location of the mass and its associated vasogenic edema, which often is of much greater volume. This chapter does not provide a detailed discussion of every possible localizing sign but rather focuses on the general principle of neurologic localization and the specific emergent syndromes that should be recognized because of their localizing and management importance (summarized in Table 4).

Disorders associated with frontal lobe lesions include early impairment of intellectual function and language function if the dominant frontal lobe is involved. Patients with bilateral frontal tumors appear to lack initiative and spontaneity—a state called abulia. Such patients may have an impaired gait with difficulty initiating walking. Personality changes include inattentiveness, apathy, and depression, as well as the less common disinhibition and inappropriate affect leading to socially inappropriate behaviors. As tumors enlarge to involve the motor cortex, contralateral motor problems, such as hemiparesis or monoparesis, may develop in these patients.

Receptive language, auditory discrimination, and memory all are important functions of the dominant temporal lobe. Patients with nondominant temporal lobe tumors may have seizures involving visual, olfactory, or gustatory hallucinations. Deep temporal tumors may cause a contralateral visual field cut (superior quadrantanopia).

The parietal lobe is demarcated form the frontal lobe by the central sulcus. Parietal lobe functions include tactile perceptions, integration of sensory, visual, and auditory information and visual discrimination in the inferior contralateral quadrant. When tumor involves the nondominant parietal lobe, inattention to the deficit (anosognosia) may be a prominent feature of the presentation.

Tumors in the occipital lobes cause a contralateral quadrantic or hemianopic defect, sometimes with sparing of central macular vision.

Tumors of the brainstem cause a great many focal deficits early in their course. Typically, a combination of cranial nerve palsies and long tract signs such as hemianesthesia or hemiparesis coupled with ataxia reflecting cerebellar involvement give a clue to the localization. Patients with brainstem tumors experience difficulty with swallowing and speech articulation. They are at risk for aspiration.

Cerebellopontine angle tumors such as acoustic neuromas impair function of the eighth cranial nerve and produce unilateral hearing loss, tinnitus, and, later, vertigo. Involvement of adjacent cranial nerves VII and V leads to facial palsy and facial anesthesia. Later cerebellar dysfunction reflects tumor growth in this area.

Tumors of the pituitary and suprasellar region produce endocrinologic abnormalities either by hormonal production by secretory adenomas or by impingement on hypothalamic-pituitary connections. Visual defects reflect chiasmatic involvement. The most common pituitary region field defect is a bitemporal hemianopia.

Pineal region tumors (germ cell tumors, pineocytoma, and pineoblastoma, as well as gliomas of this region) may compress the aqueduct of Sylvius, causing hydrocephalus. Compression of the pretectal area produces a characteristic syndrome (Parinaud's syndrome) with paralysis of upgaze, ptosis, and loss of pupillary light reflexes, along with retraction-convergence nystagmus.

Many primary tumors are capable of diffuse infiltration of the meninges. Gliomas, lymphomas, and oligodendrogliomas all may invade the subarachnoid space, producing a meningeal reaction that may mimic chronic infection. They produce variable cranial nerve and spinal root dysfunction and diffuse headache, and sometimes lead to communicating hydrocephalus. Elevated cerebrospinal fluid (CSF) protein, low CSF glucose, and positive results on cytologic studies are the diagnostic hallmarks.

On occasion, the clinician will be faced with symptoms that appear to give a clue to the patient's tumor site but in fact are false localizing symptoms. Abducens nerve (cranial nerve VI) palsies may reflect brainstem involvement but commonly are a nonlocalizing sign of raised (or, much less commonly, low) ICP. Ocular pain is of little localizing value because it may reflect any of the structures innervated by the first division of the fifth cranial nerve. Thus, eye pain may reflect any process in the anterior or middle cranial fossa. Diplopia may result from cranial nerve invasion by brainstem or leptomeningeal tumor but also can be caused by excess levels of AEDs.

Posterior head pain may reflect posterior fossa disease but also may come from the upper cervical segments, and spinal cord tumor or caudal extension of a primary brainstem tumor should be considered if the patient presents with pain in the occiput. Another sign of cervical cord disease is bilateral upper limb weakness or numbness.

Gait disorders pose another potential hazard for falsely localizing signs and symptoms. A frontal lobe gait ataxia may mimic basal ganglia or even cerebellar disease. The affected patient walks slowly, with a wide-based gait, and seems to have difficulty initiating movements. Proximal leg weakness may reflect spinal metastases from intracranial or systemic tumor, but probably the most common cause of proximal leg weakness is corticosteroid-induced myopathy.

Treatment of Brain Tumor Symptoms

Acute Raised Intracranial Pressure

The most immediate life-threatening syndromes are the herniation syndromes, but rapid increase in vasogenic edema also mandates aggressive treatment of evolving neurologic signs and symptoms. If the patient is rapidly deteriorating, intubation and hyperventilation aiming for a PCO2 of 25 to 30 mmHg are required. Mannitol is administered intravenously in a loading dose of 1 to 2 g/kg, followed by 0.5 to 1 g/kg every 6 hours as needed to control ICP. Mannitol may be transiently effective, but a rebound pressure increase sometimes develops after 24 to 48 hours. Intravenous dexamethasone in a loading dose of 20 mg followed by 6 to 10 mg every 4 to 6 hours is appropriate to start as initial treatment as well. Electrolytes and glucose levels must be monitored, and gastritis prophylaxis is required. Acute neurosurgical interventions include ICP monitoring devices and ventricular drainage or tumor decompression in cases of obstruction. Fluid management requires avoidance of hyponatremia.

Chronically Increased Intracranial Pressure

Dexamethasone in doses of 8 to 40 mg per day repairs the “leaky” blood-brain barrier of tumor vasculature. Controlling the edema helps to control headache, nausea, and seizures as well. Chronic raised ICP can cause visual loss, so careful ophthalmologic follow-up evaluation with visual field assessment is essential. Acetazolamide therapy for symptomatic plateau waves has been useful for some patients with raised ICP from intracerebral or leptomeningeal tumor.

In the doses required to treat cerebral edema, corticosteroids produce many adverse effects. These range from the easily managed gastritis symptoms and glucose intolerance to insomnia, steroid psychosis, intractable hiccoughs, and disabling myopathy. The psychosis may readily respond to reduction of steroid dose and addition of neuroleptics. Treatment of the steroid myopathy, however, will require weeks of physical therapy with attempts at steroid reduction. Anecdotal reports suggest that substitution of a nonfluorinated steroid such as methylprednisolone for fluorinated steroids such as dexamethasone may help reverse the weakness.

Many brain tumor patients remain on corticosteroids for prolonged periods. They are thus susceptible to infection, particularly with Pneumocystis jiroveci, which carries a 50% mortality rate. Among solid tumors primary and metastatic brain tumors have the highest rate of Pneumocystis infection that occur in 2% of patients and prove fatal to 40% of these. The mean duration of steroid therapy before infection was seven months with a mean dexamethasone equivalent dose of 1 to 2 mg per day. In view of these statistics chemoprophylaxis with one tablet twice daily of trimethoprim-sulfamethoxazole twice weekly should be given to patients with brain tumors who are on corticosteroids longer than 4 weeks.

With long-term use of corticosteroids, an adrenal insufficiency state may develop in these patients, manifested by lethargy, hypotension, electrolyte imbalance, and diffuse weakness on steroid withdrawal. Intercurrent systemic stressors such as surgery or systemic infection may lead to serious hypotension. Chronic glucocorticoid supplementation with hydrocortisone 10 to 20 mg per day is essential for these patients.


Neuro-oncologists often find themselves working closely with epileptologists as they seek to control seizures with minimal side effects. Because brain tumor patients frequently require corticosteroids, analgesics, anxiolytics, and antiemetics, drug-drug interactions make epilepsy management complex.

The potential interaction of AEDs with other medications required for brain tumor patients has led to reconsideration of prophylactic use of antiseizure drugs. Cytochrome P-450 enzyme-inducing drugs such as phenytoin and carbamazepine may interact with numerous chemotherapeutic agents, as well as newer biologic agents such as tyrosine kinase receptor inhibitors, reducing levels achieved and diminishing efficacy. AEDs also may increase the corticosteroid dose required for effective vasogenic edema treatment. For those patients with low-grade gliomas who may be on AEDS for many years there is also the risk of teratogenesis and osteoporosis.

A rash develops in approximately 20% of patients with gliomas treated with phenytoin or carbamazepine and cranial irradiation; the most serious and sometimes life-threatening reaction, the Stevens-Johnson syndrome, occurs in a few patients. This reaction is seen in the setting of hypersensitivity to AEDs unmasked during the taper of high-dose corticosteroids. The mechanism may be depletion of suppressor T cells by radiation, allowing emergence of the hypersensitivity syndrome. Treatment is controversial, but often the corticosteroid dose is doubled as AEDs are withdrawn abruptly.

In view of all of the potential hazards of AEDs, it is worthwhile to consider whether their prophylactic use is justified in the brain tumor population. A controlled prospective study of patients with brain tumors addressing the question of prophylaxis for epilepsy showed an overall incidence of 26% with no difference in the seizure rate between patients taking AEDS and those without prophlyaxis. A second study involving 100 patients with brain tumors confirmed the lack of efficacy of AEDs in preventing seizures or altering survival outcome in this clinical setting. These findings are consistent with those of Foy and colleagues, who conducted a prospective trial in 276 patients who had undergone craniotomy (not all of whom had tumors) who were randomized postoperatively to receive phenytoin, carbamazepine, or no treatment. No difference in the incidence of seizures (37%) or postoperative complications was found among the groups. For all of these reasons, a practice parameter published by the Quality Standards Subcommittee of the American Academy of Neurology concluded that no benefit could be established for routine prophylactic use of AEDs in patients with brain tumors.

Even after the acute phases of successful tumor treatment, epilepsy management may continue to be a major issue in quality of life for long-term survivors of low-grade brain neoplasms. A particularly dangerous late consequence of chronic antiepileptic therapy with phenytoin, carbamazepine, and phenobarbital is osteoporosis; therefore, long-term survivors should be screened regularly with dual-energy x-ray absorptiometry (DEXA) scans.

In the last 20 years, several AEDs have been introduced that offer new choices for brain tumor patients and the possibility of limiting side effects while achieving excellent seizure control. Of the newer agents, levetiracetam is emerging as an excellent choice in brain tumor patients as its level is not affected by chemotherapy drugs and it does not alter the metabolism of any chemotherapeutic agents. The clinician must be familiar with some side effects specific to these drugs (summarized in Table 5) in order to diagnose symptoms accurately, eliminate the offending drug, and avoid unnecessary diagnostic and therapeutic interventions.

Table 5. Adverse Effects of Antiepileptic Drugs: Special Issues in Patients with Brain Tumors


Potential Adverse Effect(s)


Hyponatremia (confusion, seizures)


Sedation, ataxia, weight gain


Weight gain, platelet dysfunction


Memory/word-finding problems, weight loss




Psychosis, irritability, lethargy

Deep Venous Thrombosis

Deep venous thrombosis is extremely common in patients with brain tumors, with a reported incidence of 28% to 45%. Among these patients, those with gliomas and meningiomas are at the highest risk for thromboembolism, which is a common cause of fever of unknown origin in this population. Alterations in fibrinolysis, immobility, paresis, tumor necrosis factor, steroids and neurosurgical procedures all combine to make brain tumor patients a high-risk group. Many of these patients receive vena cava filters because of the perceived risk of hemorrhage into their intracranial tumors. Retrospective studies, however, have suggested a very low risk of hemorrhage and in addition have demonstrated an incidence of complications of vena cava filters of greater than 60% in this patient group. Complications included pulmonary embolism, filter thrombosis, and postphlebitic syndrome. Low-molecular-weight heparins have a good safety profile in the neurosurgical population. Prophylaxis with enoxaparin 40 mg subcutaneously daily plus external compression stockings has been found to be superior to compression stockings alone for the prevention of venous thromboembolism after neurosurgery. Although no absolute guidelines are available, most neurosurgeons would be reluctant to institute full anticoagulation in these patients within 3 weeks of surgery. A subset of patients with late white matter changes due to radiation therapy, like patients with extensive ischemic leukoaraiosis, may be at special risk for brain hemorrhage while on heparin.


Lumbar Puncture

Lumbar punctures are not commonly employed for the diagnosis of brain tumors. With primary gliomas the yield from CSF cytology at initial diagnosis is extremely low. Leptomeningeal metasases occur in 4% to 15% of patients with solid tumors, most commonly from breast, lung, melanoma). Hematologic malignancies also commonly involve the meninges. Leptomeningeal seeding from supratentorial malignant gliomas is usually an end stage finding. Patients with leptomeningeal tumor frequently present with signs and symptoms such as focal cranial nerve neuropathies, paresis of one or more limbs, headaches, and lethargy, which would be more expeditiously evaluated with a brain MRI study.

Skull X-Ray Studies

Plain films of the skull have been used in the past to determine evidence of mass effect due to shift of the calcified pineal gland. This has been largely replaced with the modern use of computed tomography and magnetic resonance imaging. Plain radiographs of the skull are still useful for evaluating skull lesions such as dermoids and epidermoids as well as eosinophilic granulomas. These are primary lesions of the skull which frequently can be best seen with plain skull x-ray films. Epidermoid and dermoid tumors of the bone are congenital tumors derived from ectopic epithelium. These are benign lesions but can grow to be quite large and are usually painless, causing deformity of the skull. Eosinophilic granulomas may be associated with Hand-Schüller-Christian disease, a triad of diabetes insipidus, exophthalmos, and bone lesions usually found in the skull. Other primary bone lesions that can be found on plain skull films include multiple myeloma and plasmacytoma. Bone foramen abnormalities occurring in the skull include fibrous dysplasia and Paget's disease. Fibrous dysplasia can be cystic or sclerotic and usually occurs in the anterior middle fossa. This can cause proptosis of the eye and compression of the cranial nerves exiting the middle fossa causing pain and entrapment syndromes. Paget's disease is benign but thought to be a premalignant condition causing similar symptoms to fibrous dysplasia due to thickening of the bone at the skull base constricting neural foramen. Occasionally meningiomas can grow selectively into the bone rather than into the intracranial compartment. This causes thickening of the bone with compression syndromes as previously described causing proptosis of the eye, constriction of neural foramen causing cranial nerve deficits and more commonly, complaints of headache syndromes. Due to the thickening and sclerotic changes seen in the latter three disease states, these lesions are frequently evaluated with plain skull radiographs as part of the workup.

Computed Tomography

A CT scan usually is the first study obtained when a patient presents with a neurologic complaint. CT scans can be obtained quickly in most hospitals as a screening tool. Hemorrhagic lesions can be seen as a result of the increased density of blood. Infarcts will manifest acutely as edematous tissue and can be confused with tumors, especially low-grade gliomas, which may not enhance with contrast agents. Iodinated contrast agents are useful to differentiate the tumor from the surrounding edematous brain tissue, because the contrast agent is able to leak out of the vascular space into the tumor owing to breakdown of the blood-brain-barrier. The pattern of ring enhancement of tumors may be difficult to distinguish from the ring enhancement of a cerebral abscess, though brain tumors are far more common than brain abscesses in the United States. CT images have limitations at the skull base due to volume averaging artifact as the x-ray beam slides over uneven bone ridges. This makes it difficult to interpret findings in the skull base and in the posterior fossa (the region between the tentorium and the foramen magnum). CT scans are still the study of choice in patients who cannot undergo magnetic resonance imaging (MRI), such as those with implanted pacemakers, defibrillators, or other metal devices that prevent imaging by MRI.

Magnetic Resonance Imaging

MRI has become the study of choice for evaluating brain tumors. The high degree of definition of the anatomy of the brain as well as the absence of bony artifacts seen in CT scans has enabled magnetic resonance imaging (MRI) to provide exceptional images of the tumor. Localization of the lesion is precise as an MRI is done in three planes. The study of choice for evaluating brain tumors is MRI with and without contrast. Much like CT scan contrast, the MRI contrast agent gadolinium is able to penetrate into the tumor because of breakdown in the blood-brain barrier and subsequent “leak.” This enables the tumor site to become white on the T1 weighted images. Gadolinium enhancement can be crucial in identifying leptomeningeal disease as thickened areas of the dura, as well as “sugar coating” the brain itself.

Multiple sequences may be obtained from the MRI scans that highlight different properties of both the brain and the tumor. Of particular interest are the T2-weighted and fluid-attenuated inversion recovery (FLAIR) images showing the vasogenic edema created by the tumor. Evaluating this edema pattern is very important in assessing the degree of mass effect. The volume of edematous brain may represent the true limits of a glioma tumor, because tumor cells are present in the nonenhancing tissue surrounding the enhancing tumor. This provides the rationale for extending the irradiated zone in standard external beam radiation therapy to include the edematous volume of brain plus a margin of 1 to 2 cm. Presence of an edema pattern extending into the corpus callosum implies direct tumor invasion into the corpus callosum. The extremely compact fiber tracts passing through the corpus callosum will not allow edema fluid to pass through, so any edema seen is created by local tumor infiltration. This point is very important to keep in mind in assessing extent of tumor.

Endothelial proliferation with neovascularity is a hallmark of malignancy in brain neoplasms. MRI perfusion imaging can be performed in combination with conventional MRI examinations to determine the extent of neovascularity. Correlation between regional cerebral blood volume and grade of malignant glioma appears to be very good. In particular, this is especially helpful in distinguishing grade III from grade IV malignant gliomas. It has been noted that anaplastic astrocytomas with gadolinium enhancement show a higher vascularity index than that in tumors without gadolinium enhancement. Perfusion imaging may enable monitoring of the antiangiogenic effects of treatment in brain tumors. Perfusion imaging appears to be useful in assessing benign tumors such as meningiomas to determine the vascularity of these tumors, which may be associated with progression.

MRI techniques have been utilized to identify areas of brain activity in real time. This functional MRI technique has become quite useful in identifying regions of eloquent brain adjacent to tumors. By identifying language and motor areas adjacent to tumors one can determine the degree of aggressive surgical resection that can be carried out and also the best way to preserve this functional activity. Functional MRI scans are being imported into frameless stereotaxy units that are used for intraoperative surgical planning. By using discrete motor, sensory, language or visual paradigms, the relevant part of the cortical brain can be activated for identification of functional activity. The one limitation to functional MRI scanning is that it does not identify subcortical white matter tracts that are activated by the investigated activity.

A great deal of interest has centered on the capability of routine clinical MRI scanners to perform MR spectroscopy (MRS) as a technique to evaluate molecular components of a defined voxel within the brain tissue. Today's smaller voxel sizes allow a much greater degree of selectivity in evaluating components of a tumor. MRS has been applied quite extensively to malignant gliomas, for which the most important molecular components are N-acetylaspartate (NAA), creatine, choline, and a combined peak of lipids and lactate. MRS can be useful in diagnosis in that malignant gliomas exhibit a higher choline peak relative to creatine, and a lower NAA peak, with increasing grade of malignancy (Fig. 3). An additional peak beyond the NAA peak identifies lipids and lactate, which also constitute markers of malignancy and can be seen both in primary gliomas and metastatic tumors. MRS can be used to determine whether a region of enhancement in a previously treated tumor represents active tumor or radiation necrosis. MRS has been used to follow treatment effects over an extended period of time. Most useful has been the development of the multivoxel technique, in which large areas of the brain can be mapped with individual small voxels so that it is possible to determine small areas of active tumor within a large region of brain.

Figure 3.  MRI/MRS studies of WHO grade II oligoastrocytoma. Axial T1-weighted post-gadolinium MRI on day before surgery (top left panel) shows area of hypodensity that does not enhance (bottom left panel). MRS (middle panel) shows area of high choline (Cho; 1.8 times greater than adjacent brain) and low NAA levels, as well as a smaller region of no significant metabolite levels, which is presumed necrosis. The resection was limited to a region of elevated choline with no significant NAA. Cr, creatine; NAA, N-acetyl aspartate; WHO, World Health Organization.  

MRI has the capability to specifically identify vascular structures such as major arteries and veins. MR angiography (MRA) and MR venography (MRV) can be useful in evaluating major vessels at the base of the skull for skull base tumors. The quality of MRA and MRV has been steadily improving, and both of these techniques provide a noninvasive means for evaluating these vascular structures without the need for angiography.

Positron Emission Tomography

PET imaging uses a radioactive isotope, 18F-fluorodeoxyglucose (FDG), to image metabolism of glucose in the brain. Because glucose is the sole fuel for brain tissue this metabolic imaging allows visualization of brain physiology. PET scans will be positive in glioblastoma, primary CNS lymphoma, oligodendroglioma, and malignant meningioma owing to the high uptake of FDG. Anaplastic astrocytomas and oligodendrogliomas, meningiomas, and metastatic tumors show variable FDG uptake; low-grade gliomas and radiation necrosis show little to no FDG uptake (Figs. 4 and 5).

Figure 4.  Positron emission tomography (PET) scan of recurrent glioma. The patient had undergone surgical resection of an anaplastic oligodendroglioma 15 months earlier. Areas of contrast enhancement on T1-weighted magnetic resonance imaging (MRI) scan (left panel) have high fluorodeoxyglucose (FDG) accumulation, consistent with recurrent high-grade tumor on the co-registered FDG-PET image (right panel).  

Figure 5.  Positron emission tomography/magnetic resonance imaging (PET/MRI) studies showing radiation necrosis of the brain. The patient had had a recurrence of a basal cell carcinoma involving the skin overlying the zygoma and the left periauricular region, with extension into the temporalis muscle. He underwent surgical resection, but because of perineural invasion, he received radiation therapy at a dose of 6400 cGy postoperatively to the tumor bed. Approximately 2 years later, he presented with aphasia and confusion. A, T1-weighted post-gadolinium MRI showed a ring-enhancing mass in left temporal lobe, a finding highly suggestive of malignancy. B, FLAIR MRI showed extensive edema. C, An 18F-fluorodeoxyglucose (FDG) scan, however, showed patchy areas of hypometabolism within the anterior and lateral aspects of the left temporal lobe—findings consistent with radiation effect, not malignancy. The patient underwent surgery with resection of the lesion, which confirmed it to be radiation necrosis without any evidence of malignancy.

Intraoperative Ultrasound Examination

In general, brain tumors are echogenic, so real-time ultrasound studies can be used for localization. Ultrasound examination can be used to identify normal intracranial structures such as the ventricles, the falx, and the tentorium. Exact tumor localization is necessary because of the limited operative exposure to the brain. The ultrasound transducer can be applied to the dura surrounding the brain or placed directly on the brain tissue. Ultrasound examination can be quite useful in finding large brain tumor cysts, which can cause herniation of the brain during surgery. With accurate localization of the cyst by this means, a needle can be inserted to drain the cyst, thereby rapidly decompressing the brain.


The successful resection of a brain tumor requires removing the tumor only, without injuring surrounding normal brain. The approach to removal of brain tumors follows the mantra of real estate agents: “location, location, location.” Deep tumors that are in noneloquent regions of brain can be easily accessed and removed, whereas superficial tumors may not be easily resectable owing to their location within extremely eloquent brain tissue.

The goals of surgery are (1) to establish the histology of the lesion; this is frequently better performed through a craniotomy with a more aggressive debulking of the tumor, rather than a simple stereotactic biopsy, so that there is no question of sampling error; (2) to debulk the mass effect of the tumor to correct a neurologic deficit and to prevent imminent death in patients with large tumors and early herniation syndromes; and (3) to debulk the tumor to increase efficacy of radiation therapy and chemotherapy, which produce the best response rate when they are used with minimal tumor burden.

Establishing a tissue diagnosis is extremely important for therapeutic and prognostic considerations. Frequently, the histologic diagnosis may be straightforward; however, a category of tumors exists that manifest as mixed gliomas, in which tissue sampling can induce errors in diagnosis. A limited biopsy specimen may show a solitary cell type, but further examination of a more extensive specimen may reveal the presence of other components. This is particularly important when the other components identified may change the histopathologic interpretation from a grade II glioma to an anaplastic grade III glioma. Noninvasive means of establishing tissue diagnosis have been pursued, primarily through the use of MRS. Significant progress has been made to determine histologic diagnosis of tumors with MRS; however, MRS is still not specific enough to form the basis for major therapeutic decisions. Histopathologic determination remains the gold standard.

The most common presenting manifestations in patients with glioma are headache and seizures. Debulking large tumors will reduce the dural stretch, decreasing headaches. The incidence of seizures in patients with malignant gliomas can be decreased by at least 75% with attempts at gross total resection. Recovery from other neurologic deficits manifesting as hemiparesis, visual field loss, or aphasia, for example, will depend on whether the impaired brain tissue is simply compressed by the tumor mass or whether the tumor itself has directly destroyed these neural tracts. In the former case, but not the latter, debulking the tumor will help relieve the patient's symptoms.

As a result of liberal use of CT and MRI scans, herniation syndromes with brain tumors, as noted earlier under Pathophysiology of Signs and Symptoms (in the clinical presentation section), are uncommon presentations. Occasionally, however, patients present with central or uncal herniation syndromes from late-stage brain tumors. These patients need to have an emergency CT or MRI scan and undergo rapid evaluation for possible surgery for reduction of mass effect as a life-saving measure.

Another rationale for maximal debulking is to reduce the tumor burden by one or two log orders of cells. Such cytoreduction is perfomed in the hope of increasing efficacy of both radiation therapy and chemotherapy for these tumors.


Radiation is commonly used to treat many different types of brain tumors. The radiation oncologist must decide on many factors in the treatment plan for an individual patient, including treatment volume, dose, and fractionation. Treatment techniques including three-dimensional conformal therapy, stereotactic radiotherapy, and intensity-modulated radiation therapy (IMRT) are being used in a majority of patients presenting with brain tumors. Proton therapy, which has increasing availability in the United States, may be used for specific cases, especially for tumors at the skull base.

Radiation Therapy: Technical Details

Treatment planning starts with a review of the MRI scan to identify the target volumes. The appropriate volume to be irradiated varies according to the tumor type. Most brain tumors including gliomas and meningiomas are treated with focal radiation to the lesion plus a margin. The actual abnormality seen on imaging studies is termed the gross tumor volume (GTV). Additional tissue surrounding the GTV that is thought to potentially contain tumor cells is included in the clinical target volume (CTV). For some tumors such as glioblastomas, which can be highly infiltrative, the CTV includes the area of edema as demonstrated on the FLAIR or T2 sequences plus a 1.5- to 2.0-cm margin. For other tumors that do not infiltrate, such as meningiomas, pituitary adenomas, acoustic neuromas, and craniopharyngiomas, the CTV typically would be much tighter than that used for glioblastomas. An additional volume is included around the CTV to take into account day-to-day setup error and to allow for buildup of dose, resulting in the patient treatment volume (PTV). Treatment planning proceeds by placing the patient on a CT simulator, generally in the supine position. Multiple immobilization devices are available, including thermoplastic masks, bite blocks, and systems that include motion tracking devices. After the CT scan is obtained, a previously obtained MRI scan can be fused onto the treatment planning system to better define the target volumes. In addition to the tumor, normal structures including the optic nerves, chiasm, cochlea, and brainstem often are drawn as areas to limit the radiation dose. Computerized treatment planning is then used to generate radiation beams and dose distributions.

Most brain tumors are treated with focal radiation therapy; however, for some tumors, such as PNETs and metastatic germ cell tumors of the CNS, it may be necessary to include the entire craniospinal axis in the irradiated zone. In such cases, a special technique must be used. For craniospinal irradiation, the optimal patient position is prone. This facilitates daily setup and treatment, with the spinal canal included in one or two fields entering from the patient's back and the entire cranial contents included in opposed lateral fields. Because abutting fields are used to encompass the entire craniospinal axis, the possibility exists for overlap between fields, resulting in a potential overdosage to part of the spinal cord. Therefore, great care must be taken during the setup of these fields. In many institutions, a technique termed feathering is used in which the matchlines between fields are changed periodically to ensure that an overdose or underdose does not occur in the same region of the spinal cord throughout the entire treatment.

Stereotactic Radiotherapy

Stereotactic radiotherapy is a technique for delivering high-dose radiation to a target volume with very tight margins, thereby sparing surrounding normal tissues. This technique can be implemented in a number of ways using various devices, including (1) a Gamma Knife machine, which contains 201 fixed cobalt sources that converge to a single point; (2) a conventional linear accelerator (“linac”) outfitted with additional hardware so that it can deliver focused radiation, generally using between 3 and 5 arcs; and (3) delivery of charged particles such as protons, which, because of their physical properties, deposit energy in a narrower region than is possible with x-rays.

Stereotactic radiation can be delivered in a single large fraction. When this is done, the technique is termed stereotactic radiosurgery. Despite of its name, no surgery is performed. A sterotactic headframe often is screwed to the skull. In adults, this can be done using local anesthesia, but in children, conscious sedation or general anesthesia must be used. The headframe allows for localization of the tumor in a three-dimensional coordinate system and also immobilizes the head during treatment, thereby allowing for delivery of radiation with precision to a very tightly defined volume. For linear accelerator-based radiosurgery, the patient has a CT scan performed with the headframe in place. At our institution, MRI with gadolinium contrast is performed before placement of the headframe. The MRI images are then fused to the CT images, and a treatment plan is devised that targets the lesion while minimizing dose to normal structures. For Gamma Knife radiosurgery, an MRI-compatible headframe is placed before a contrast-enhanced MRI scan is obtained for use in target definition. The doses of radiation that are used in stereotactic radiosurgery range from 10 to 24 Gy and are based on the histologic tumor type and the volume of tissue irradiated. The doses used are based on the likelihood of tumor control and the risk of developing radiation necrosis. The choice between a Gamma Knife and a linear accelerator is based on availability. Generally, Gamma Knife doses are prescribed to the 50% isodose line, and those with linear accelerator-based therapy, to the 80% isodose line.

Stereotactic radiotherapy also can be delivered in a fractionated regimen over several days or weeks. Of course, in this instance a headframe that is screwed into the skull cannot be used. Therefore, relocatable headframes have been developed. These headframes use custom-made moldings conforming to the patient's occiput and a bite block to maintain a precise and reproducible fit day to day.

A large experience has been accumulated with the use of stereotactic radiotherapy, in either a single fraction or a fractionated regimen, for the treatment of intracranial metastases, arteriovenous malformations, and primary brain tumors. The use of this modality to treat arteriovenous malformations is beyond the scope of this chapter, and its use in treating intracranial metastases is covered in Chapter 56 . The use of stereotactic radiotherapy techniques for the treatment of meningiomas, pituitary adenomas, and acoustic neuromas is discussed in greater detail in the corresponding sections in this chapter.

Delivery of charged particles including protons has been used to treat brain tumors. Because of their physical properties, protons result in more limited dose to normal tissues than is possible with conventional x-rays. Only six centers in the United States have protons or other charged particles available for medical treatment, although a number of new facilities have recently been opened, and additional sites are in the planning phase.

Intensity-Modulated Radiation Therapy

The technique of IMRT uses multiple segmented fields with inverse treatment planning algorithms. In addition to defining the doses to be delivered to the GTV, CTV, and PTV, dose constraints are placed on normal structures to limit their radiation exposure. Constraints include the maximum dose to the entire structure and limits to a portion of the structure, as well as the maximum point dose within the structure. The most common mechanism of delivering this dose is use of multiple fields, typically 5 to 9, with use of a multileaf collimator to divide each field into multiple beamlets. From 50 to more than 100 beamlets may be used. Most modern radiation departments have access to this technique.

Adverse Effects after Irradiation of the Brain or Spine

Acute and Early Delayed Effects after Cranial Irradiation

A variety of adverse effects may follow irradiation of the CNS. These effects can be divided into acute, which occur during the treatment; delayed early, which occur within a few months of radiation; and late, which occur months to years later.

Acute effects of cranial irradiation include fatigue, nausea, headaches, anorexia, and alopecia. Patients often complain of fatigue, not just with cranial irradiation, but with irradiation to other regions of the body, and as a result they often need to sleep longer than usual or take naps during the day. Patients may experience nausea within hours after the administration of the radiation and headaches during the course of treatment. Nausea usually is well controlled with antiemetics such as ondansetron or granisetron. Nausea and headaches are thought to be caused by radiation-induced edema and can be ameliorated with corticosteroids. In patients receiving craniospinal radiation, the nausea and anorexia may be compounded by the radiation that the upper gastrointestinal tract receives as an exit dose from the spinal field. Hair loss generally starts after the scalp has received 20 to 30 Gy. It generally is not permanent, but regrowth of hair may take months, and the new hair often is thinner than the original or even of a different color. In areas that receive a high dose of radiation, especially with tangentially directed fields, alopecia may be permanent. Other acute side effects of cranial irradiation may include accumulation of cerumen in the ear canals and serous otitis media.

The most common delayed early effect from radiation is the somnolence syndrome which is characterized by excessive drowsiness, nausea, and irritability. If it occurs, it generally does so one to three months after radiation has been completed. This syndrome is thought to be due to transient, diffuse demyelination. It usually is seen after whole-brain irradiation for ALL but also can develop after limited-field irradiation for brain tumors. The syndrome resolves spontaneously, but steroids can shorten its duration. Delayed early effects occurring after cranial irradiation also can take the form of focal neurologic signs due to intralesional reactions related to tumor response or perilesional reactions related to edema or demyelination.

Brain Necrosis and Neurocognitive Deficits Following Cranial Irradiation

The pathophysiology of late effects from CNS irradiation is poorly understood. Some of the effects may be caused by degenerative changes in the supporting glial cells, whereas others may be caused by vascular changes due to endothelial cell loss and capillary occlusion.

One of the most serious late effects of cranial irradiation is brain necrosis, which may cause significant and persistent neurologic injury. Second, it may produce progressive cerebral edema and mass effect requiring prolonged corticosteroid use or surgery. Third, it may be confused with tumor growth, resulting in the inappropriate use of antitumor therapies. The onset of radiation necrosis includes behavioral changes—lethargy and dementia, headache and papilledema, and seizure. Clinical signs and symptoms are usually identified from 2 to 3 years after irradiation, although confirmed cases been detected as early as 9 months and as late as 16 years after completion of radiation therapy. The signs and symptoms are strongly related to the site of radiation necrosis; however, the most common clinical signs are focal motor deficits. Radionecrosis often is difficult to distinguish from recurrent tumor by CT or MRI, which may show increased signal intensity on T2-weighted images and contrast enhancement on T1-weighted images. As discussed earlier (in the “Diagnostic Imaging” section), and as shown in Figure 5 , however, PET scanning with FDG may help distinguish viable tumor from necrotic tissue. Surgical exploration often is necessary not only to establish a diagnosis but also as a therapeutic intervention to remove the region of necrosis. Histopathologically, the changes seen in radiation-induced necrosis generally are limited to the white matter and include focal coagulative necrosis and demyelination. Accurate data regarding the incidence of brain necrosis based on CT and MRI findings come from a randomized trial of irradiation for low-grade gliomas. In this study, the 2-year actuarial incidence of brain necrosis was 2.5% for patients receiving 50.4 Gy and 5% for those receiving 64.8 Gy.

Neurocognitive deficits often are observed after cranial irradiation, especially in young children. Many of the sequelae appear several years after treatment of children with brain tumors, which mandates long-term follow-up. Sequential assessments of neurocognitive function demonstrated progressive deterioration during 6 years after radiotherapy in children with ALL treated with 18 Gy of whole-brain irradiation. Numerous studies of neurocognitive function in children after whole-brain irradiation for ALL have been performed. In summary, these findings show that whole-brain irradiation can lead to decline in neurocognitive function, an effect that appears to be greater in younger children and with higher doses of radiation (24 Gy) but can be seen after 18 Gy. It also is possible that an interaction between methotrexate (intrathecal or high-dose systemic) and whole brain irradiation causes these late effects.

Another treatment complication associated with cranial radiotherapy is leukoencephalopathy. This most is often associated with intravenous or intrathecal methotrexate and cranial irradiation. Young age also is an important risk factor; however, leukoencephalopathy can affect all age groups. Histologically, multifocal white matter destruction with loss of myelin can be seen, especially in the periventricular regions; MRI scans show these periventricular abnormalities. CT scans also may reveal the presence of intracerebral microcalcifications due to mineralizing microangiopathy. The clinical expression of leukoencephalopathy ranges from mild evidence of white matter injury on neuroimaging studies to severe necrotizing leukoencephalopathy with profound neurologic impairment and, in some cases, death. Mild or subclinical cases are more common than severe necrotizing leukoencephalopathy. The bulk of the experience comes from children who received 24 Gy of whole-brain radiation along with high doses of intravenous and intrathecal methotrexate. The frequency of leukoencephalopathy is low in patients who receive cranial radiotherapy and intrathecal methotrexate or cranial irradiation and intravenous methotrexate but may be as high as 45% in patients who receive all three treatments. In general, methotrexate is most toxic when given during or after radiation.

Although the major portion of the literature on neurocognitive function after cranial irradiation concerns children, some data on adults are available. Taphoorn and colleagues found no significant differences in neurocognitive function between patients with low-grade gliomas who received radiation (45 to 63 Gy) and those who did not. In another retrospective study, young adults with low-grade brain tumors treated with 54 to 56 Gy of radiation to limited fields often showed a transient, early delayed drop in neuropsychological performance at 6 months; however, the risk of long-term cognitive dysfunction was low, at least up to the age of 4 years.

In an NCCTG randomized study of 64.8 versus 50.4 Gy for treatment of low-grade gliomas, data regarding cognitive performance were collected prospectively. Analysis of these data with a median follow-up time of 7.4 years in survivors showed that the vast majority of patients with normal baseline findings on the Mini-Mental Status Examination (MMSE) maintained these after radiotherapy. Patients with abnormal MMSE results before radiotherapy were more likely to have an improvement in cognitive abilities than deterioration after receiving radiotherapy. Armstrong and coworkers conducted prospective, comprehensive, longitudinal neuropsychological testing on 26 adult patients with low-grade supratentorial brain tumors, mostly gliomas, who had received radiotherapy. Nine patients underwent testing at 6 years after radiotherapy. No declines were noted on most neurocognitive tests. Seven of the 37 neuropsychological tests showed improvement over 6 years. However, declines emerged only after 5 years on selected tests of cognitive function such as visual memory.

On the basis of these and other studies, one recent review on the neurocognitive effects of radiotherapy on patients with low-grade gliomas came to the conclusion that the weight of evidence suggests only sporadic, limited neurocognitive damage from focal radiotherapy at the doses usually prescribed. These patients do not appear to suffer from widespread cognitive impairment or dementia.

Cranial irradiation can result in arterial vascular problems such as vessel obliteration or narrowing, resulting in a stroke-like syndrome. These complications are rare, but when they occur they are more likely to happen after irradiation of the parasellar region.

Endocrine Deficits after Cranial or Spinal Irradiation

Endocrine problems are common after cranial irradiation, particularly in children. Such problems include growth hormone deficiency and thyroid and gonadal dysfunction.

The hypothalamus is more radiosensitive than the pituitary gland and is responsible for endocrine dysfunction at lower doses. At higher doses (greater than 40 Gy), however, both the anterior pituitary gland and the hypothalamus contribute to endocrine dysfunction. Of all the hormones, growth hormone is the most likely to show deficiency following irradiation. Growth hormone deficiency is seen in a majority of children who have received whole-brain irradiation. One study found that in children with ALL given 24 Gy to the whole brain, 56% developed growth hormone deficiency, whereas no such problems were seen in children given 18 Gy, at least 4 years later. The latency of onset is dose-dependent, being shorter with higher doses. Growth may be further impaired by spinal irradiation, which directly affects vertebral body growth center.

Precocious puberty may also occur after relatively low doses, on the order of 18 Gy. Deficiencies in the other hormones, such as gonadotropins, TSH, and ACTH, are rare after doses below 40 Gy, however. Thyroid dysfunction is common in patients with brain tumors treated with high-dose radiotherapy. Constine and associates studied endocrine function in 32 patients with brain tumors not involving the hypothalamic-pituitary region who received 39.6 to 70.2 Gy to this region. Hypothalamic or pituitary hypothyroidism developed in 65% of the patients. Fourteen of 23 (61%) postpubertal patients had evidence of hypogonadism as manifested by oligomenorrhea or low estradiol levels or low testosterone levels. Half of the patients had mild hyperprolactinemia. Subtle abnormalities in adrenal function were seen in 35% of patients.

In patients receiving craniospinal radiation, hypothyroidism also may occur secondary to exit dose to the thyroid gland. In a study of radiation therapy for brain tumors not involving the hypothalamic-pituitary axis from the Christie Hospital in Manchester, the incidence of hypothyroidism was 15% or 33% (P = 0.013), respectively, for patients receiving cranial or craniospinal irradiation. The mean spinal dose was 29 Gy, and the exit dose to the thyroid gland ranged from 10 to 15 Gy.

Optic Neuropathy after Cranial Irradiation

Irradiation of tumors that are close to the optic nerves or chiasm may result in sufficient dose to these structures that optic neuropathy is a concern. Two major classes of optic neuropathy are recognized: anterior optic neuropathy and retrobulbar optic neuropathy. The former is thought to be due to vascular injury affecting the nerve head inside the globe anterior or adjacent to the lamina cribrosa. This is associated with swelling of the optic head, in contrast with retrobulbar optic neuropathy, which is due to more proximal injury to the optic nerve. Diagnostic criteria for retrobulbar optic neuropathy include (1) visual loss (monocular or binocular) accompanied by corresponding visual field defects, (2) funduscopic examination often showing a pale optic disc but without edema, (3) onset 6 months to several years after radiation therapy that delivered a significant dose to the optic nerve-chiasm, and (4) no radiologic evidence of visual pathway compression. MRI scans may show pathologic contrast enhancement of the region of the optic nerve-chiasm that received a high dose of radiation.

Parsons and colleagues examined radiation-induced optic neuropathy in patients receiving radiation treatment for primary extracranial head and neck tumors at the University of Florida. Of 215 optic nerves at risk, these investigators found anterior optic neuropathy in 5 nerves and retrobulbar optic neuropathy in 12. No injuries were observed in 106 optic nerves that received less than 59 Gy. The 15-year actuarial risk of optic nerve neuropathy after 60 Gy more was 11% when daily fractions less than 1.9 Gy were used, versus 47% when 1.9 Gy or more was used.

The foregoing data suggest that the optic nerve-chiasm tolerance is at least 59 Gy; however, the University of Florida population did not include patients with intracranial tumors compressing the optic nerve-chiasm. It is possible in the latter situation that the tolerance of the optic nerve-chiasm is lower as a result of ischemic injury. In some patients with pituitary adenomas or craniopharyngiomas treated with irradiation, optic neuropathy developed after doses as low as 45 to 50 Gy, although in many of these cases the daily fraction size was greater than 2 Gy. On the basis of these results, most investigators currently recommend limiting the optic chiasm-nerve dose to 50 Gy in 1.8- to 2-Gy fractions in the treatment of pituitary adenomas or craniopharyngiomas. For other brain tumors requiring higher doses, most radiation oncologists would try to restrict the optic nerve-chiasm dose to 54 Gy or lower. Adherence to these guidelines should keep the risk of radiation-induced optic neuropathy extremely low (1% or less) but unfortunately will not completely eliminate it.

Second Malignant Neoplasms Developing after Cranial Irradiation

Second malignant neoplasms including malignant gliomas and meningiomas remain relatively uncommon consequences of radiation therapy for brain tumors. Recent reports raise concerns that the addition of adjuvant chemotherapy may increase the risk of second malignancies in long-term survivors of childhood brain tumors. This may be especially true after prolonged use of alkylating agents and etoposide with or without irradiation.

Myelopathy after Spinal Irradiation

A delayed early effect that can be seen after irradiation of the cervical spine is Lhermitte's syndrome, which is characterized by an electric shock-like sensation precipitated by forward neck flexion. The symptoms typically start weeks to a few months after completion of radiation therapy. They are maximal at first but abate with time, without the development of any objective signs. The paresthesias most commonly occur in the lumbosacral region but also can involve the upper and lower extremities and the upper back. This transient form of Lhermitte's syndrome can occur after doses of radiation well within accepted spinal cord tolerance and is not associated with any permanent late sequelae. The pathogenesis is thought to involve an inhibitory effect on oligodendrocytes that results in transient reversible demyelination. A more ominous form of Lhermitte's syndrome can manifest after a longer latency period after completion of radiation therapy (at least a year), which then progresses to chronic radiation myelopathy.

Chronic myelopathy is the most catastrophic late effect that can occur after spinal cord irradiation. Nearly half of the affected patients will die from complications. A biphasic distribution in the latency period from irradiation to the onset of myelopathy has been observed. The early peak is from 12 to 14 months, and the second peak, from 24 to 28 months. The pathologic insults that lead to myelopathy include damage to oligodendroglial cells, causing demyelination and white matter necrosis, and death of endothelial cells, resulting in vascular injury. In some patients, partial neurologic dysfunction develops; others progress to complete paraplegia or quadriplegia. No clinical or radiologic findings are pathognomonic for radiation-induced myelopathy. Therefore, the diagnosis usually is made by a combination of (1) neurologic abnormalities corresponding to a level just below the irradiated region, (2) a history of spinal cord irradiation with a high total dose (greater than 45 Gy) at least 6 months before the onset of signs and symptoms, (3) MRI findings of increased signal intensity on T2-weighted images in the irradiated region, and (4) exclusion of other etiologic factors. The MRI findings in spinal cord myelopathy can mimic tumor recurrence with gadolinium enhancement on MRI (Fig. 6).

Figure 6.  Magnetic resonance imaging (MRI) in a patient with radiation necrosis of the spinal cord. The patient received 45 Gy of radiation to the cervical cord (C1 to C6) for grade II astrocytoma of the cord. Four years later she presented with weakness of the upper extremities. T1-weighted post-gadolinium MR image showed enlargement of the cervical cord and enhancement (arrow), thought to be consistent with recurrence. The patient received chemotherapy without improvement and eventually died as a result of respiratory failure. Autopsy of the spinal cord showed chronic radiation changes and necrosis of the cervical cord but no evidence of recurrent tumor.

The probability of developing chronic radiation myelopathy is dependent not only on the total dose delivered to the spinal cord but also on the fraction size. It is now well recognized that larger fraction sizes are associated with more severe later effects. This correlation was not always appreciated, however, and many of the cases of chronic myelopathy described in the literature occurred in patients who received 40 to 60 Gy to the cord in large daily fractions (2.45 to 5 Gy). Using standard fraction sizes (1.8 to 2 Gy per day), a commonly observed limit for dose to the spinal cord is 45 Gy. In a study of the incidence of myleitis after irradiation of the cervical cord, Marcus and Million found that of 1112 patients, only 2 (0.18%) developed chronic radiation myelopathy. Two out of 471 patients (0.42%) receiving between 45 and 50 Gy developed this complication. Myelitis developed in none of the 442 patients who received between 40 and 45 Gy and none of the 75 who received greater than 50 Gys. The investigators’ conclusion was that even with doses of up to 50 to 55 Gy to the cervical cord, the likelihood of developing chronic myelopathy was extremely low. Even if the spinal cord dose is limited to 45 Gy, myelitis may still develop. This report cited published cases in which myelitis developed with doses less than 45 Gy or even less than 40 Gy, given in 1.8- to 2-Gy daily fractions. Such cases, however, are extremely rare, with an estimated risk of 0.2% or less for the development of chronic myelopathy at this dose.


Chemotherapy of brain tumors involves many of the same problems as chemotherapy for systemic cancer, including lack of specificity, intrinsic or developing cellular resistance, intolerance of normal tissue to drug toxicity, synergistic toxicity between chemotherapy and radiation therapy, and systemic toxicity. Brain tumor therapy also is associated with specific problems of drug delivery across the blood-brain or blood-tumor barrier. Cerebral edema may impede drug delivery, and corticosteroids that effectively treat the edema may “repair” tumor vessels and impede the delivery of chemotherapy to the tumor. Thus, a recurring disappointment in clinical chemotherapy trials has been the failure to translate promising preclinical chemotherapy findings into meaningful improvement in patient survival.

Further complicating assessment of chemotherapy in brain tumor patients is the observation that corticosteroids also often reduce MRI contrast enhancement and relieve symptoms, making it difficult to distinguish chemotherapy effect from corticosteroid effect. Further confusion may arise when neuroimaging documents enlargement of mass after stereotactic radiosurgery that may progress over many months. Because PET scanning may not reliably distinguish radiation necrosis from tumor progression, the results of concurrent chemotherapy may be ambiguous. Conventional MRI response criterion of 25% to 50% reduction in contrast-enhancing tumor, therefore, may not always be a true measurement of chemotherapeutic efficacy. Radiation therapy also may result in endovascular changes, making access to the brain more difficult for chemotherapeutic agents. Attempts to circumvent this problem by administering chemotherapy before irradiation remain an area of active investigation.

Clinical criteria for chemotherapeutic success are similarly confusing. Performance status, measured in many clinical trials by the Karnofsky Performance Scale (KPS), is affected by tumor site, presence of seizures, and adverse effects of AEDS and steroids. Thus, appropriate measurement of chemotherapeutic efficacy should not be median survival alone. Median progression-free survival or 6-month progression-free survival and reduction in seizure frequency, corticosteroid requirement, and focal deficits all are parameters that must be measured. For long-term survivors, cognitive impairment and other serious acute and chronic neurologic toxicities of cytotoxic chemotherapy and newer molecular strategies factor into the assessment of chemotherapeutic program efficacy.

Finally, chemotherapy of primary brain tumors for the past 20-some years has been designed on the assumption that the problem is one of local control, with few recurrences outside the original tumor site. With longer survivals and better local tumor control, however, however, a higher rate of multicentric cerebral and even leptomeningeal disseminated disease has been noted among patients with high-grade glial tumors, and oncologists may need to design future systemic regimens with these considerations in mind.

An important consideration in the chemotherapy of brain tumors is the ability of the drugs to cross the blood-brain barrier. This physiologic barrier defines the restricted transport between blood and the CNS of water-soluble, ionized molecules larger than about 200 daltons. The blood-brain barrier is formed by the endothelial cells of brain capillaries, with some contribution from astrocytes. Brain capillaries differ from capillaries elsewhere in the body by the presence of tight intercellular junctions. The brain's extracellular or intersititial fluid is an ultrafiltrate essentially identical to CSF. Capillaries of the choroid plexuses are fenestrated, allowing access of large protein molecules into the plexus stroma. The epithelial cells separating stroma from CSF, however, have apical tight junctions. The brain lacks a lymphatic system.

These features of the blood-brain barrier and the blood-CSF barrier exclude entry of large molecules such as proteins and limit entry of smaller molecules to their ability to cross the lipid bilayer of cells. Lipid-soluble molecules such as nicotine, ethanol, heroin, and the alkylating agent BCNU (carmustine) pass readily across the blood-brain barrier. Another function of the blood-brain barrier may be that of pumping out chemicals potentially harmful to the brain. P glycoprotein, a blood-brain barrier protein, is present in the membrane of some brain tumor cells and serves to reduce the intracellular concentration of several chemotherapeutic agents. The P glycoprotein is encoded by a gene responsible for a form of resistance to chemotherapy: the MDR1 gene (for multidrug resistance). Numerous studies have linked the expression of P glycoprotein with resistance to topoisomerase inhibitors including VP-16 (etoposide), VM-26 (teniposide), camptothecin, and other drugs such as paclitaxel and their derivatives. In preclinical studies in nude mice bearing human glioblastoma implanted tumors, coadministration of P glycoprotein inhibitors such as valspodar (SDZ PSC-833) resulted in marked improvement in access of paclitaxel and 90% reduction in tumor volume. Not all malignant gliomas contain tumor cells that express P glycoprotein, and the expression in such cells does not correlate with tumor grade. Little evidence links the expression of MDR1 gene to patient response to specific chemotherapeutic drugs.

Another mechanism involved in glioma cell BCNU resistance is O[6]-methylguanine DNA-methyltransferase (MGMT) expression. MGMT repairs nitrosourea-induced DNA damage by catalyzing the transfer of a methyl group from the O[6] position of guanine to its own molecule through a cysteine acceptor site. Not all BCNU-resistant glioma cell lines overexpress this enzyme, but the presence of MGMT recently has been shown to correlate with survival in patients with glioblastomas treated with nitrosoureas. Patients who had high levels of MGMT had a median survival period of 9 months, versus 15 months for those with low levels of MGMT. Administration of MGMT inhibitors resulted in increasing the cytotoxic effects of BCNU.

Quantitative examination of the blood-brain barrier and the brain-tumor barrier has resulted in new avenues of investigation for brain tumor chemotherapy. Methotrexate entry into brain tumors can be enhanced by intra-arterial delivery, but the use of hyperosmolar mannitol for this purpose results in a far greater increase of drug entry into normal brain tissue than into tumor. Human and rat studies of BCNU, cisplatin, and other agents after blood-brain barrier disruption in both primary glial tumors and primary CNS lymphoma have demonstrated significant toxicity. Attempts to deliver BCNU and other drugs intra-arterially have been largely abandoned because of unfavorable survival compared with that for patients who received intravenous treatment; the poor outcomes were due in large part to marked cerebral and ocular toxicity.

The blood-brain barrier in brain tumors is impaired in many pathologic states, including traumatic injuries, infections, ischemia, and neoplasms. In brain tumors, the degree to which the tight endothelial cell junctions of the blood-brain barrier are disrupted and vascular permeability is thereby increased varies within even a single tumor. The persistence of a relatively intact barrier impedes entry of some water-soluble chemotherapeutic agents into areas of tumor, leading investigators to look for methods of opening the blood-brain barrier that are less toxic than intra-arterial mannitol. Another method of opening the barrier has been with the agent RMP-7, a bradykinin analog. On the basis of earlier promising studies in rat tumors in which intravenous RMP-7 selectively increased uptake of carboplatin into tumors, a randomized controlled trial of intravenous carboplatin and RMP-7 versus carboplatin and placebo was conducted in patients with recurrent malignant glioma. No significant differences were found in median survival times, median time to progression, or findings on neuropsychological, functional independence, or quality of life assessments. The use of RMP-7 had no effect on the pharmacokinetics or toxicity of carboplatin.


Clinical Considerations

Patients with supratentorial gliomas may present with general signs and symptoms such as changes in mental status, seizures, headaches, papilledema, and nausea and vomiting, as listed earlier under Clinical Presentation. They also may have focal signs and symptoms, dependent on tumor location, as discussed previously.

The past few decades have seen a change in clinical presentation with supratentorial low-grade astrocytomas. In the pre-CT era, patients often presented with headaches, papilledema, and motor weakness. In more recent series, however, in which CT or MRI scans were performed routinely, at least two thirds of patients presented with seizures, and very few had other signs or symptoms. A possible explanation is that in the pre-CT era, many of these patients would have been placed on antiseizure medication and would not have been diagnosed with a brain tumor until their tumor grew large enough to cause signs and symptoms of increased ICP.

A history of seizures also is extremely common in patients with oligodendrogliomas, occurring in 70% to 90% of cases. The duration of symptoms can be prolonged. In a study by Ludwig and coworkers, 55% of patients had symptoms for longer than 1 year before diagnosis, and 37% had symptoms for longer than 3 years. A few patients had symptoms for 10 to 15 years before diagnosis.

Low-grade astrocytomas decrease in frequency with increasing age. Their incidence is highest between the ages of 20 and 40 years and decreases in patients older than 50. Oligodendrogliomas display a similar age-related frequency. Conversely, high-grade astrocytomas, including glioblastomas, increase in frequency with increasing age (after the age of 60 years; see Fig. 1).

Pathologic Classification of Supratentorial Gliomas

Astrocytomas arise from astrocytes, the supporting cells of the central nervous system. Glial fibrillary acidic protein (GFAP) is expressed in the cytoplasmic processes that extend from the astrocytes; therefore, antibodies against this protein can be used in immunohistochemical studies. Over the years, many different histopathologic staging systems have been used. Kernohan devised a four-tiered system that classified tumors into grade 1, the slowest-growing tumors, through grade 4, the most malignant tumors. A three-tiered system including astrocytoma, anaplastic astrocytoma, and glioblastoma, developed by Ringertz, was used in many cooperative trials.

More recently, Daumas-Duport and Scheithauer, reviewing cases from the Mayo Clinic and Sainte-Anne's Hospital in Paris, devised a system using the presence of nuclear atypia, mitoses, endothelial proliferation, and necrosis to grade tumors from 1 through 4. This system led to a better discrimination of outcome for patients who underwent treatment at the Mayo Clinic than did the Kernohan system.

The most widely used system today for classification of astrocytomas is the WHO system. Grade I is reserved for pilocytic tumors, which are the most benign in terms of histology, generally curable with surgery alone. Because they are more common in children than in adults, they are discussed in greater detail later under Childhood Brain Tumors. The remaining categories are grade II (diffuse astrocytomas), grade III (anaplastic astrocytomas), and grade IV (glioblastomas).

Diffuse astrocytomas (grade II) are poorly defined, gray tumors that expand the parenchyma and obliterate normal gray matter–white matter boundaries (Fig. 7). Microscopically there are increased numbers of irregularly distributed astrocytes with mildly atypical nuclei (Fig. 8A). Tumor cells can be seen infiltrating into normal brain tissue at some distance from normal tissue. Diffuse astrocytomas often are classifed into one of three subtypes: fibrillary, which is the most common subtype, protoplasmic, or gemistocytic. Fibrillary astrocytomas are composed of tightly interlacing bundles of small, spindle-shaped cells amid a predominantly fibrillar matrix. Gemistocytic astrocytomas contain plump cells with distinct, round pink cytoplasm arranged on a more delicately interlacing fibrillar matrix, whereas protoplasmic astrocytomas are composed of small, round, regular cells with indistinct cytoplasmic boundaries arranged on a loosely fibrillar stroma.

Figure 7.  Diffuse low-grade astrocytoma, gross specimen. WHO grade II astrocytoma arises diffusely and infiltrates the right frontal lobe. Gross determination of the tumor's boundaries is almost impossible, but the tumor is evident as an ill-defined area of enlargement, with loss of distinction between the gray and white matter.  

Figure 8.  Histologic appearance of low-grade astrocytoma versus anaplastic astrocytoma. A, Low-grade astrocytoma (WHO grade II) shows low cellularity, slight nuclear pleomorphism, no endothelial proliferation, and no necrosis. B, Anaplastic astrocytoma (WHO grade III) shows increased cellularity, pleomorphism, and mitotic activity compared with the grade II tumor in A. Anaplastic astrocytoma also lacks endothelial proliferation and necrosis.  

Microscopically, anaplastic astrocytomas (WHO grade III) are distinguished from grade II tumors by their greater cellularity, increased nuclear pleomorphism, and of most importance, mitotic activity (see Fig. 8B). In both these and grade IV glioblastomas, tumor cells can be seen infiltrating into surrounding normal brain tissue, often at a great distance from the primary tumor mass. Histologically, glioblastomas are distinguished from anaplastic astrocytomas by the presence of endothelial proliferation or necrosis (Fig. 9). The necrosis can form a serpentine pattern referred to as “pseudopalisading,” in which tumor cells crowd around the edges of the necrotic region. The presence of necrosis is of particular importance in grading gliomas and has been associated with shorter survival. Grade IV glioblastomas show areas of hemorrhage, necrosis, and cystic change on gross examination (Fig. 10). Glioblastomas can manifest with multicentric disease, but this occurs in less than 5% of cases as determined at autopsy of patients with untreated tumors.

Figure 9.  Glioblastoma (grade IV astrocytoma): histologic appearance. A, Pseudopalisading cells surrounding regions of necrosis. B, Endothelial proliferation, which in this case has reached dramatic proportions, with the formation of tangled clusters of neovascular channels, often referred to as “glomeruloid” blood vessels because of their resemblence to renal glomeruli.  

Figure 10.  Glioblastoma: gross specimen. The tumor appears as a necrotic, hemorrhagic infiltrating mass.  

In addition to pathologic grading by morphology, techniques have been developed to measure cell kinetics based on the idea that faster-growing tumors are more malignant. These include immunohistochemical staining using antibodies directed against Ki-67 (MIB-1) or proliferating cell nuclear antigen (PCNA). Another technique involves calculation of an S phase fraction by measuring the incorporation of bromodeoxyuridine (BUdR) or iododeoxyuridine (IUdR) into tumor cells after intravenous injection of the agent. The predictive power of all three of these techniques has been compared with that for clinical parameters. Problems with application of these methods remain, so they are not routinely used, but they offer some promise as a means of providing prognostic information independent of histology. Likewise, it is very possible that immunohistochemical stains for genetic changes found in glial tumors also will be used in the future to assess prognosis.

The different grades of astrocytoma carry very different prognoses. With WHO grade I (pilocytic) astrocytomas, a cure rate greater than 90% after surgical resection alone can be expected. By contrast, with WHO grade IV tumors, the median survival period is 1 to 2 years, even after aggressive combined-modality therapy. With WHO grade III tumors, long-term survival rates on the order of 20% are usual, but the median survival period is approximately 2 years. WHO grade II tumors carry a better prognosis than grade III tumors; nevertheless, recurrence with a higher-grade tumor is common with grade II tumors, although this may take 6 to 8 years to occur.

Most oligodendrogliomas occur in the cerebral hemispheres (80%), but they can occur in the lateral and third ventricles. Oligodendrogliomas arise from oligodendrocytes, which produce myelin. They often involve the subcortical white matter, with extension into the cerebral cortex. On gross examination, oligodendrogliomas often are soft and gelatinous and better circumscribed than astrocytomas. They frequently contain calcifications. Despite their gross appearance suggesting that they are contained, they can infiltrate surrounding tissues, including the subarachnoid space and leptomeninges,

Owing to fixation artifact, oligodendrogliomas appear microscopically as sheets of cells with nuclei surrounded by a clear halo of cytoplasm, giving the cells the appearance of a “fried egg” (Fig. 11A). Unlike astrocytomas, oligodendrogliomas lack fibrillary cytoplasmic processes. Calcifications are present in 90% of these tumors. A common finding is a network of capillaries, lending a “chicken wire” pattern. The current WHO classification for oligodendrogliomas is a two-tiered system of grade II (oligodendroglioma) and grade III (anaplastic oligodendroglioma). Features suggestive of anaplasia include cytologic atypia and increased mitotic activity (Fig. 11B). Two other features, microvascular proliferation and pseudopalisading necrosis, when present in an oligodendroglial tumor would classify it as a grade III tumor, in contrast with an astrocytic tumor, in which these features would make it a glioblastoma (grade IV). The prognosis with anaplastic oligodendrogliomas is less favorable than with low-grade tumors, but they are much more sensitive to chemotherapy than their astrocytoma counterparts, as discussed in detail later on. In the Mayo Clinic series, the 5- and 10-year survival rates and median survival period for low-grade oligodendrogliomas were 75%, 46%, and 9.8 years, respectively, versus 41%, 20%, and 3.9 years for high-grade tumors.

Figure 11.  Histologic appearance of oligodendroglioma versus anaplastic oligodendroglioma. A, In WHO grade II oligodendroglioma, oliogodendrocytes are uniform cells with small, round nuclei and a characteristic perinuclear halo (“fried egg” cells). B, Grade III oligodendroglioma shows cytologic atypia and increased mitotic activity compared with the tumor in A.  

Distinct from oligodendrogliomas are mixed gliomas or oligoastrocytomas. According to the WHO classification, this latter category includes tumors that show “a conspicuous mixture of two distinct neoplastic cell types resembling the tumor cells in oligodendroglioma and diffuse astrocytoma.” The two components may be in different regions or diffusely mixed together. This definition is vague in that an oligodendroglioma with a very small astrocytic component may be classified by some pathologists as a pure oligodendroglioma but by others as a mixed oligoastrocytoma. In the WHO classification, these tumors are subdivided into oligoastrocytomas (grade II) and anaplastic oligoastrocytomas (grade III). Grading of these tumors is very difficult because the two components often differ in grade. In particular, it is problematic how to classify a tumor that has oligodendroglioma-like regions in a background that otherwise appears to be a glioblastoma. Some experts have referred to these as glioblasto mas with oligodendroglioma component, which may carry a better prognosis than that for ordinary glioblastomas.

Overall, patients with oligodendrogliomas have a better prognosis than patients with astrocytomas. Oligoastrocytomas might be expected to have an intermediate prognosis, and in some series this is the case. For example, in the Mayo Clinic experience, the 5- and 10-year survival rates and the median survival period were 46%, 17%, and 4.7 years for low-grade diffuse astrocytomas, compared with 73%, 49%, and 9.8 years for low-grade oligodendrogliomas. The respective figures for patients with oligoastrocytomas were in between, at 63%, 33%, and 7.1 years. In the University of California at San Francisco series, however, 5- and 10-year survival rates for patients with pure oligodendrogliomas and with oligoastrocytomas were identical. The discrepancy between these series may be due to the difficulty in categorizing these tumors, and to interinstitutional differences in pathologic interpretation.

Imaging of Supratentorial Gliomas

WHO grade I (pilocytic) astrocytomas show enhancement on CT and MRI scans. By contrast, WHO grade II astrocytomas typically are poorly defined, hypodense or isodense lesions on CT scans that do not enhance. Either localized or homogeneous enhancement, however, can be seen in up to 30% to 40% of cases, with calcification in 5% to 10% of cases. MRI shows low signal intensity on T1-weighted images and high signal on T2-weighted images (Fig. 12A and B). Grade III anaplastic astrocytomas and grade IV glioblastomas generally enhance with contrast although in one study, enhancement was not seen in 40% of the former and in 4% of the latter (Fig. 12C and D). The region of enhancement typically has a ring-like appearance surrounding an area of necrosis. The perimeter of the enhancing region does not define the border of the tumor, and malignant cells may be present beyond this. T2-weighted MRI images show abnormalities that are more extensive than those seen on a contrast-enhanced CT scan or T1-weighted MRI scan. In one study in which stereotactic biopsy findings were correlated with radiologic findings in patients with gliomas, isolated tumor cells often were found as far away as the region showing increased signal intensity on T2-weighted images.

Figure 12.  Magnetic resonance imaging: low-grade astrocytoma versus glioblastoma. A, WHO grade II astrocytoma: T1-weighted post-gadolinium axial image shows no enhancing mass; however, a region of hypodensity is seen. B, Corresponding FLAIR image of the tumor in A shows abnormality consistent with edema. C, WHO grade IV glioblastoma: T1-weighted post-gadolinium axial image shows the presence of a large mass compressing the right lateral ventricle. A rim of enhancement with central necrosis is typical of these tumors. D, Corresponding FLAIR image of the tumor in C shows extensive peritumoral edema. FLAIR, fluid-attenuated inversion recovery; WHO, World Health Organization.

At least 50% of oligodendrogliomas show calcifications, which can be appreciated on plain films of the skull as well as on CT scans. Enhancement of oligodendrogliomas can be seen on CT and MRI scans, but this often is mild and poorly defined.

Genetics of Supratentorial Gliomas

Genetic Changes in Astrocytomas

The p53 pathway frequently is disrupted in all grades of astrocytomas. This can occur by mutation of p53 itself, by overexpression of MDM2, which degrades p53, or by mutation or deletion of p14ARF, which positively regulates p53 by inhibiting MDM2 expression. In one study, the p53 pathway was disrupted by one of these mechanisms in 67% of diffuse astrocytomas, 72% of anaplastic astrocytomas, and 76% of glioblastomas. Therefore, disruption of the p53 pathway appears to be an early event in the genesis of astrocytomas. It is possible that mutation of p53 in low-grade astrocytomas leads to genomic instability that sets the stage for additional mutations that lead to the formation of higher-grade tumors. The p53 gene (TP53) is located on the long arm of chromosome 17 (17p), which frequently is lost in both low-grade and high-grade astrocytomas, suggesting that p53 is the target for this genetic change.

The Rb pathway, which regulates the G1/S transition, commonly is disrupted in high-grade astrocytomas. This can occur through deletion or mutation of RB itself; deletion or mutation of CDKN2A, which encodes the cyclin-dependent kinase inhibitor p16; deletion or mutation of CDKN2B, which encodes the cyclin dependent kinase inhibitor p15; or amplification of the CDK4 gene. Ichimura and coworkers noted that one of these changes had occurred in 67% of glioblastomas and 21% of anaplastic astrocytomas, but in none of 15 low-grade astrocytomas. These findings suggest that progression to a higher grade glioma is facilitated by deregulation of the G1/S transition.

Loss of heterogosity (LOH) of 13q and 9p has been identified in high-grade astrocytomas, and these deletions may lead to alterations in the Rb pathway. LOH of chromosome 13 occurs in 30% to 40% of higher-grade astrocytomas. The Rb gene (RB), which maps to 13q14, is the likely target of this deletion. CDKN2A is located on 9p21, and this chromosomal region is homozygously deleted in at least 30% to 40% of glioblastomas and in approximately 10% of anaplastic astrocytomas, but never in diffuse astrocytomas. Because two different gene products, CdkN2A and p14ARF, both are transcribed from a single locus on 9p21, deletion of this region leads to simultaneous disruption of both the Rb and p53 pathways.

Astrocytic tumors also frequently display high-level expression of PDGF ligands and receptors. Most studies have found that the PDFG-α receptor is overexpressed in 60% to 90% of low- and high-grade astrocytomas. By contrast, overexpression of the ligand PDGF-A or -B is not very common in low-grade gliomas but is seen in higher-grade tumors, suggesting that this growth factor participates in an autocrine loop in these tumors.

Genetic Changes in Glioblastomas

Approximately 80% of all glioblastomas develop without any prior history of a glioma and are termed primary glioblastomas. The clinical history usually is short, on the order of months, and the patients tend to be older, with a median age of 55 when diagnosed. These tumors tend to be large with lots of edema, central necrosis, and ring enhancement. In 20% of the cases, the patient has an antecedent history of a lower-grade glioma (grade II or III), generally 5 to 10 years previously. These secondary glioblastomas tends to arise in younger patients, with a median age of 40 years.

The genetic changes seen in primary and secondary glioblastomas are very different (Fig. 13). Primary glioblastomas often show homozygous deletions of the CDKN2A/p14ARF locus on 9p21 (in 30% to 40% of cases) or amplification of CDK4 and MDM2 on 12q13–15 (in 8% to 13% of cases). Either of these sets of genetic changes will cause disruption of both the p53 and Rb pathways.

Figure 13.  Molecular alterations during “glioma-genesis.” Two pathways for the development of glioblastomas have been described. In the secondary pathway shown, the patient has a history of a prior lower-grade astrocytoma, whereas in the primary pathway, no such history exists. The Rb/p16/Cdk4 pathway is disrupted in primary and secondary glioblastomas to a similar extent with LOH 13q/RB mutation and CDK4 amplification. However, 9p/CDKN2 deletion, another means of disrupting the Rb pathway, is seen more commonly in primary glioblastomas. These three changes targeting the Rb pathway appear to be mutually exclusive. EGFR, epidermal growth factor receptor; LOH, loss of heterozygosity; PDGFR-α, platelet-derived growth factor receptor-a.

EGFR amplification, leading to overexpression, is seen in approximately 35% of primary glioblastomas, rarely in anaplastic astrocytomas, and never in diffuse astrocytomas. Another 15% of primary glioblastomas overexpress EGFR without amplification. Approximately half of glioblastomas with EGFR amplification express a mutant form of the receptor. The most common mutant is known as ΔEGFR or EGFRvIII, which is missing exons 2 to 7, resulting in an in-frame deletion of 801 base pairs of the coding sequence of the extracellular domain. This particular mutant form of EGFR is constitutively activated and cannot be downregulated. Its expression in glioblastoma cells has been associated with increased proliferation, decreased apoptosis, and increased tumorigenicity and invasion in vivo. Even in glioblastomas that overexpress wild-type EGFR, which is not constitutively active, the ligands EGF and TGF-α often are coexpressed, which may activate an autocrine loop that allows for self-stimulation.

Up to 90% of primary glioblastomas exhibit deletion of 10q, on which is located the tumor suppressor gene PTEN. The retained PTEN allele is mutated in approximately 50% of tumors with 10q loss, leading to complete loss of functional PTEN protein. Therefore, the incidence of PTEN mutation in glioblastoma is at least 45%. One study suggested a potential association between lower PTEN levels and shorter survival in patients with glioblastomas, although this did not reach statistical significance. PTEN mutations have been found in anaplastic astrocytomas, although at a much lower frequency than in glioblastomas. This study found the presence of PTEN mutations in anaplastic astrocytoma to be a powerful prognostic factor portending a poorer outcome.

In contrast with primary tumors, secondary glioblastomas rarely display EGFR amplification. Secondary glioblastomas have loss of 10q, but they do not have mutations of the retained PTEN gene. Both the p53 and Rb pathways generally are disrupted in secondary glioblastomas, as is the case with primary glioblastomas; however, the mechanisms tend to be different. Both wild-type p53 alleles are lost in more than half of all secondary glioblastomas, generally one allele by deletion and the second by mutation. By contrast, p53 mutations rarely are seen in primary glioblastomas (occurring in less than 10% of cases). The Rb pathway is disrupted in secondary glioblastomas, often through loss of both wild-type alleles. Secondary glioblastomas may also show disruption of the p53 and Rb pathways through promoter methylation of CdkN2A and p14ARF.

Genetic Changes in Oligodendrogliomas and Oligoastrocytomas

Two chromosomal abnormalities frequently are seen in oligodendrogliomas: deletion of 1p and deletion of 19q. Oligodendrogliomas exhibit LOH at 19q in up to 88% of cases and LOH at 1p in up to 100% of cases. LOH of 1p and 19q is common in both grade II and III oligodendrogliomas, so these changes probably occur early during tumorigenesis. By comparative genome hybridization, loss of 1p and 19q is found in 79% and 74%, respectively, of oligodendrogliomas but in only 25% and 30% of oligoastrocytomas. Finer mapping indicates that the regions of interest lie in 1p36 and 19q13, although the specific genes have not yet been identified. Some evidence suggests that AOs with 1p or 19q deletions respond much more favorably to chemotherapy than tumors without these losses. A correlation also has been found between oligodendroglioma location and the presence of these genetic alterations. Anaplastic oligodendrogliomas arising from frontal, parietal, or occipital lobe were more likely to contain 1p or 19q deletions than those arising in temporal lobe, insula, or diencephalon.

Although 1p and 19q losses are the most common genetic abnormalities in oligodendrogliomas, other mutations have been reported. In one genetic analysis of 446 CNS tumors, TP53 mutations and homozygous deletions of CDKN2 were identified in 5% and 11%, respectively, of oligodendrogliomas. The same mutations were seen in 37% and 15%, respectively, of oligoastrocytomas.

Surgery for Supratentorial Gliomas: Extent of Surgical Resection

The role of extent of surgical resection in the treatment of gliomas has been controversial for many years. Virtually all of the reviews in this subject area have been retrospective; therefore, they are subject to significant selection bias. It is likely that patients with the most favorable, easily resectable lesions within noneloquent brain regions have tended to undergo aggressive resections, whereas those with deeper lesions in more eloquent brain regions have been selected for stereotactic biopsy to minimize surgical morbidity. Some experts have advocated maximal resection for gliomas, reasoning that this approach is associated with improved survival. A possible explanation for this outcome may be that fewer cells are left behind that need to be eradicated with irradiation and chemotherapy.

Other investigators have advocated stereotactic biopsy alone as the preferred mode of histological diagnosis in the treatment of patients with gliomas. Stereotactic biopsy affords a histological diagnosis with a low complication rate (2% to 5%). However, stereotactic biopsy may lead to an inaccurate pathologic diagnosis. In a review of data for 81 consecutive patients initially diagnosed with stereotactic biopsy who subsequently underwent a craniotomy, Jackson and colleagues showed that the histopathologic classification based on stereotactic biopsy findings was incorrect in 38% of the patients. This represented sampling error in these patients, 96% of whom underwent biopsy at outside institutions. Despite the fact that these were tumors located in eloquent brain and previously treated by neurosurgeons, gross total resection could be achieved in 57% of these patients once they reached a major tertiary care hospital (M.D. Anderson Cancer Center, Houston, Texas). Major complications were seen in only 12.3% of the patients undergoing aggressive craniotomy for resection of the malignant glioma located in eloquent brain.

To specifically address the role of aggressive resection of gliomas in prolonging survival, Lacroix and coworkers performed a retrospective multivariate analysis of outcomes for 416 patients using prospectively collected data. Postoperative volumetric MRI scans were obtained on all patients so that a blinded neuroradiologist was able to determine extent of resection given as a percentage of the original tumor. Patients with a resection of 98% or more of the tumor volume had a median survival period of 13 months, versus 8.8 months for patients with a less than 98% resection. Also found to be significant were age, KPS score, and extent of tumor necrosis. Even when statistical analysis controlled for these other three factors, extent of resection remained a significant variable.

Navigation during Surgery

Intraoperative MRI scanners have allowed for real-time evaluation of the extent of tumor resection. Imaging of relevant anatomy can be performed during surgery with MRI in order to provide real-time feedback regarding the surgical resection. Schulder and colleagues reviewed the cases of 93 patients who underwent neurosurgery aided by a low-field intraoperative MRI scanner. These investigators found that surgery was directly affected by imaging in 51% of the operations. Second lesions not otherwise evident on the initial preoperative scan were seen in 21 patients, and in another 14 patients unnecessary dissection was avoided as a result of real-time imaging. Intraoperative MRI scanners with field strengths of 1.5 T are being used not only for anatomic imaging but also for perfusion imaging and MR angiographic imaging. Frameless stereotactic units have been available for many years, allowing for downloading preoperative MRI scans to an intraoperative computer workstation to create a three-dimensional reconstruction of the images. The patient's head can be directly referenced to the images for precise intraoperative localization to less than 2 mm. Investigators have been able to import both PET and functional MRI data into the frameless stereotaxy units to map out functional brain during surgery.

Intraoperative cortical and subcortical mapping has been used to define eloquent brain during resection of gliomas. Intraoperative cortical mapping has enabled neurosurgeons to become more aggressive with tumors located within eloquent brain regions. This application is especially useful with low-grade gliomas that occur immediately adjacent to motor and speech areas. Cortical mapping may be done with the patient lightly anesthetized but awake enough to respond to questions so that speech areas can be mapped.

Complications of Surgery

A recent analysis by the Glioma Outcome Project reviewed perioperative complications and neurologic outcomes for patients who underwent craniotomy for the diagnosis and treatment of gliomas. The Glioma Outcome Project was a prospectively compiled database capturing information on 788 patients. Of these, 499 underwent either a first or second craniotomy for treatment of their malignant glioma, and the remaining 289 patients underwent stereotactic biopsy only. No difference was found in the characteristics of the patients who underwent first or second craniotomy. Analysis of the perioperative symptoms, however, revealed that patients undergoing a second craniotomy had a higher incidence of altered level of consciousness and papilledema, whereas those undergoing a first craniotomy had a higher incidence of headache. The incidence of depression was higher in patients undergoing a second craniotomy: 11% in the group undergoing the first craniotomy versus 20% in the patients undergoing a second craniotomy. This difference may reflect the presence of a chronic disease state. The rate of systemic infections also was higher in patients undergoing a second craniotomy at 4.4%, versus zero in patients undergoing a first craniotomy. This is not unexpected, because these patients frequently have been heavily pretreated with radiation therapy and chemotherapy, as well as long-term steroid use, causing significant immunosuppression. Evaluation of postoperative neurologic status showed that neurologic status was the same or better in 92% of the patients undergoing their first craniotomy, whereas this rate dropped to 82% in patients undergoing their second craniotomy. Nevertheless, the overwhelming majority of patients are benefited by a debulking procedure to reduce neurologic deficits. As has been shown in multiple other studies, the most important preoperative factor associated with good neurologic outcome has been KPS score. Those patients with higher KPS scores fared better with surgery than did patients with lower scores. This finding is a reflection of degree of neurologic injury. The Glioma Outcome Project is perhaps the only prospectively collected database for evaluating outcomes of surgery in the treatment of malignant gliomas. The analysis of these data reveals that the incidence of further neurologic deficit with craniotomy is only 8% in patients undergoing their first operation, versus 18% in patients undergoing a second craniotomy for resection of a malignant glioma. In view of the severity of the disease, this is certainly an acceptable risk. Of greater importance, they are objective prospectively obtained data showing improvement in neurologic outcome with aggressive debulking surgery. An important point is that these data were compiled from participating institutions, and the decision for surgery was left to the discretion of the surgeon; therefore, certainly selection bias obtained in terms of who underwent radical surgery versus stereotactic biopsy. The investigators do not discuss the frequency of tumors in eloquent versus noneloquent brain. Nevertheless, findings of this study are representative of the general practice of surgical neuro-oncology throughout the nation.

Convection-Enhanced Delivery

Local therapy for gliomas has been an intriguing concept because of the anatomically limited nature of the disease. Gliomas do not metastasize outside of the CNS and frequently recur within a 2-cm margin of the previous resection. Simple diffusion will work for small molecules and some forms of chemotherapy. With the advent of biologic modifying agents with large molecular weights, however, another development has been convection-enhanced delivery—that is, pressure-driven delivery of drugs directly into the brain tissue, causing a gradient to induce bulk flow into the interstitial space of the brain, thereby distributing macromolecules across a distance measured in centimeters, rather than millimeters. Use of this technique to deliver two such agents has been described. The first agent is a conjugate of transferrin and a mutant Diphtheria toxin infused into patients with malignant gliomas. The investigators reported at least a 50% reduction in tumor volume in 9 of 15 patients. The second clinical trial administered a conjugate of IL4 with Pseudomonas exotoxin for patients with recurrent malignant gliomas. This study was a safety and toxicity trial and has been carried forward using IL-13 into the recently concluded Phase III Randomized Evaluation of CED (Convection Enhanced Delivery) of IL13-PE Compared to Gliadel Wafer with Survival Endpoint Trial (also known as the PRECISE Trial). The survival data for this study are not available at this time. The preclinical and some of the clinical studies using convection-enhanced delivery show excellent distribution through out the brain, and the agent has been reported to pass through normal brain to reach a second site of tumor, with excellent drug delivery to the second tumor site. Box 1 summarizes the recommended approach to management of supratentorial gliomas.

Box 1.


  • Grade I (pilocytic) astrocytomas: Surgery is curative. If residual tumor is seen on postoperative imaging, the patient should undergo a second craniotomy to resect the entire tumor. Radiation therapy and chemotherapy have limited usefulness for treatment of these tumors.
  • Grade II (low-grade) astrocytoma: Surgery is the mainstay of therapy for tumors in noneloquent regions of brain. In patients younger than 40 years of age who undergo gross total resection, no additional therapy is given. In patients younger than 40 with incomplete resection and patients older than 40 with or without complete resection, adjunctive treatment with radiation therapy (54 to 60 Gy) is indicated.
  • Grade III astrocytoma (anaplastic astrocytoma) and grade IV gliomas (glioblastoma): Surgery is required to establish tissue diagnosis, preferably with debulking as well. Chemotherapy is begun with temozolomide during radiation therapy and continued for six cycles after completion of the radiation regimen. The radiation dose generally is 60 Gy. Tumor tissue is sent for analysis of O[6]-methylguanine DNA-methyltransferase (MGMT) promoter activity. Temozolomide is offered to all patients, however, regardless of promoter methylation status.

Radiation Therapy for Supratentorial Gliomas

Radiation Therapy for Low-Grade Gliomas

Numerous retrospective reports have been published on the use of radiation therapy for low-grade gliomas. These studies are plagued with a myriad of problems, making them inconclusive. In order to contain sufficient patient numbers, most of these reports span decades, sometimes going back as far as the 1940s. Obviously, before the advent of CT scans, accurate treatment planning would have been difficult, as would good radiologic follow-up to document relapses. Before the 1960s, radiation was given using orthovoltage machines, which do not have the capability to deliver treatment to deep tissues. The doses given to some of the patients would be considered inadequate by today's standards. In these studies, immediate postoperative radiation often was given to patients with poor prognostic features, thereby potentially skewing the results in favor of observation.

The results of these retrospective studies have largely been superseded by the results of three randomized trials, two from the European Organization for Research and Treatment of Cancer (EORTC) and one from the North Central Cancer Treatment Group (NCCG; Table 6 ). The EORTC 22844 trial randomized adults with supratentorial low-grade astrocytomas, oligodendrogliomas, or mixed oligoastrocytomas to receive either 45 Gy or 59.4 Gy of radiation after surgery. The NCCG study was similar except that patients were randomized to receive either 50.4 or 68.4 Gy after surgery. Patients underwent a range of surgical procedures, including biopsy and subtotal or gross total resection. Neither study showed any benefit to the higher dose in terms of overall survival or progression-free survival. If anything, the NCCG study showed a worse survival in patients receiving the higher dose of radiation (64.8 Gy). The 5-year survival rate in both studies ranged from 58% to 72%.

Table 6. Results of Treatment for Low-Grade Gliomas: Selected Randomized Trials



Histologic Type

Treatment Protocol

No. of Patients

5-Year Survival

5-Year PFS

Karim et al


9% A, grade 1

45 Gy


58%, P = 0.94

47%, P = 0.73

 EORTC 22844


60% A, grade 2

59.4 Gy






22% O







9% mixed





van den Bent et al


2% A, grade 1



66%, P = 0.87

35%, P < 0.0001

 EORTC 22845


60% A, grade 2

54 Gy






25% O







10% mixed





Shaw et al NCCTG


32% A or mixed (A > O)

50.4 Gy


72%, P = 0.48

55%, P = 0.65



68% O or mixed (O > A)

64.8 Gy




A, astrocytoma; EORTC, European Organization for Research and Treatment of Cancer; NCCTG, North Central Cancer Treatment Group; O, oligodendroglioma; PFS, progression-free survival.

The EORTC also performed study 22845, in which patients who underwent surgical treatment for astrocytomas, oligodendrogliomas, or mixed oligoastrocytomas were randomized to receive either adjunctive irradiation in a dose of 54 Gy or no upfront radiation therapy. Patients in the latter group had the option of receiving radiation if progression of disease was observed. No difference was seen in overall survival between the early radiotherapy and observation groups (see Table 6: 5-year survival rate, 68% versus 66%; median survival period, 7.4 versus 7.2 years). A statistically significant advantage, however, was observed in the patients who received radiation in terms of progression-free survival (55% versus 35% at 5 years; median time to progression, 5.3 versus 3.4 years; P < 0.0001). The absence of any difference in survival has supported the position of experts who advocate delaying radiation. Conversely, those favoring upfront radiation in the treatment of low-grade gliomas have cited the improved progression-free survival in the irradiated group, and the fact that at 1 year, seizures were better controlled in the radiation treatment arm.

These randomized trials confirmed the importance of certain prognostic factors in low-grade gliomas. In the NCCG trial, three prognostic factors—histologic subtype, patient age, and tumor size—were consistently associated with overall survival in multivariate analysis. The 5-year survival rates for patients with oligodendroglioma-predominant tumors and those with astrocytomas were 74% and 56%, respectively (P = 0.0001); for patients younger than 40 years of age and those ≥40 or older, 77% and 60%, respectively (P = 0.025); and for preoperative tumor size less than 5 cm and 5 cm or greater, 81% and 61%, respectively (P = 0.008). In the EORTC 22844 study, the extent of resection had a great impact on overall survival: Patients who had greater than 90% of their tumor removed did much better than those who underwent a biopsy. Multivariate analysis of data from the two EORTC trials confirmed that astrocytoma histology, age older than 40 years, and preoperative size greater than 6 cm all were unfavorable prognostic factors but also uncovered a few others, such as tumor crossing midline and the presence of neurologic deficits before surgery. On the basis of these randomized trials, it is evident that in adults with low-grade gliomas, no difference in survival is achieved whether radiation therapy is given postoperatively or delayed until recurrence. Therefore, a reasonable management strategy would be observation in a patient with a completely resected low-grade glioma. Many investigators, however, still advocate radiation after complete resection in older patients (older than 40 years) or after incomplete resection. The standard dose is 54 Gy given in 1.8-Gy fractions over 6 weeks. It is common to treat the MRI-defined gross tumor volume (GTV) with a 1.5- to 2-cm margin.

Radiation Therapy for High-Grade Gliomas

The earliest randomized trial to demonstrate that radiation therapy was beneficial in the treatment of high-grade gliomas was conducted by the Brain Tumor Study Group (BTSG 69-01) (Table 7). Patients were randomly assigned to one of four groups: supportive care, whole-brain radiation therapy (WBRT), BCNU, or WBRT plus BCNU. The role of BCNU is discussed further under Chemotherapy for Gliomas later on. This BTGS trial clearly showed a benefit for radiation alone versus supportive care with a median survival for 9 months versus 3.5 months and a 1-year survival rate of 24% versus 3% (P = 0.001). On the basis of this trial, radiation therapy has remained an essential component in the treatment of high-grade gliomas. A second randomized trial from the Scandinavian Glioblastoma Study Group confirmed the efficacy of radiation using a lower dose of WBRT (45 Gy) compared with supportive care (Table 7).

Table 7. Results of Treatment for High-Grade Astrocytomas: Selected Randomized Trials Focusing on Radiation and Radiation Modifiers



Gliomas (%)

Treatment Protocol

No. of Patients

Median Survival (mo)

18-Month Survival (%)


Walker et al



Supportive care




RT superior to supportive care (P = 0.001)

 BTSG 69-01











RT (60 Gy WBRT)








RT (60 Gy WBRT) + BCNU




RT + BCNU superior to supportive care (P = 0.001)

Kristiansen et al


Supportive care




RT superior to supportive care




RT (45 Gy WBRT)








RT (45 Gy WBRT) + Bleo





Bleehen and Stenning


RT (45 Gy)



60 Gy superior to 45 Gy (P = 0.04)




RT (60 Gy)




Chang et al



RT (60 Gy WBRT)




No difference between any groups

 RTOG 74-01



RT (60 Gy WBRT) + 10 Gy boost




No improvement with 70 Gy




RT (60 Gy WBRT) + BCNU













Curran et al



RT (60 Gy) + BCNU




No improvement with HFX

 RTOG 90-06



RT (72 Gy HFX) + BCNU




For patients <50 yr, median survival better with standard fractionation

Selker et al



125I implant (60 Gy) +




No improvement with implant


(10% AA)


BCNU + RT[*] (60.2 Gy)





 BTCG 87-01



RT[*] (60.2 Gy) + BCNU





Laperriere et al



RT (50 Gy) + BCNU




No improvement with implant




RT (50 Gy) + BCNU + 125I implant (60 Gy)





Nelson et al







No improvement with misonidazole

 RTOG 79-18



RT + Miso + BCNU





Prados et al



RT (60 Gy) + PCV



Preliminary: no improvement with BUdR

 RTOG 94-04



RT (60 Gy) + BUdR + PCV




Souhami et al



RT (60 Gy) + BCNU

186 (total in both arms)


22 (2 yr)

Preliminary; no improvement with radiosurgery boost

 RTOG 93-05



Radiosurgery boost + RT (60 Gy) + BCNU



18 (2 yr)


AA, anaplastic astrocytoma; BCNU, carmustine; Bleo, bleomycin; BTCG, Brain Tumor Cooperative Group; BUdR, bromodeoxyuridine; DTIC, dacarbazine; HFX, hyperfractionation; MeCCNU, sem ustine; Miso, misonidazole; PCV, procarbazine, cisplatin, vincristine; RT, radiation therapy; RTOG, Radiation Therapy Oncology Group; SGSG, Scandinavian Glioblastoma Study Group; WBRT, whole-brain radiation therapy

* During the early part of the study, external beam radiation was given to whole brain to a dose of 43 Gy, followed by a boost to the tumor volume for an additional 17.2 Gy. From May 1989, WBRT was dropped, and the entire dose was restricted to tumor volume.

All patients in this trial had AAs.

The results from BTSG 69-01 and two successive BTSG studies were pooled together to examine dose response. The original studies were all randomized, but patients were not randomly assigned to different radiation doses; therefore, the analysis of dose was a retrospective one. With this caveat kept in mind, the study showed a benefit to using 60 Gy versus 50 Gy (median survival, 10.5 versus 7 months; P = 0.004). The Medical Research Council in Great Britain performed a randomized study examining dose in radiation treatment for high-grade astrocytomas. After surgery patients were randomly assigned to receive either 45 or 60 Gy of radiation. The treatment volumes used in this study were very generous but less than for whole-brain levels. Among patients receiving 45 Gy, irradiation was mostly to the entire supratentorial region. Among those who received the higher dose, a dose of 40 Gy was delivered to volumes similar to those in the first group, followed by an additional 20 Gy to the tumor volume with a 1-cm margin.

The current standard of care in RTOG studies is to define the initial volume as the preoperative lesion with edema as seen on T2-weighted MR images with a 2-cm margin. This volume is treated to 46 Gy, followed by a boost to the gadolinium-enhancing lesion seen on T1-weighted images with a 2.5-cm margin to 60 Gy.

By pooling the results of three RTOG studies and using a nonparametric recursive partitioning technique, Curran and associates identified several significant prognostic factors for survival in patients with high-grade gliomas: histologic type, KPS score, age, neurologic function, and duration of symptoms. Patients could be placed into groups I through IV with different outcomes on the basis of these factors.

Data have been accumulated regarding the radiologic response of high-grade astrocytomas to radiation therapy. In one multicenter trial, patients had CT scans done at several points in management: (1) preoperatively, (2) at the end of radiation therapy, (3) 6 to 8 weeks after the end of radiation therapy coinciding with the start of chemotherapy, (4) 8 weeks after the first course of chemotherapy, and then (5) every 3 to 4 months. The radiation dose consisted of 44 Gy to the whole brain, followed by 14 Gy to the tumor volume. Twenty-two of 63 evaluable tumors (35%) responded to irradiation, defined as showing a decrease in the enhancing tumor volume by 25% or more. The vast majority of responding tumors, 20, showed a response by the end of radiation therapy. In two tumors, the response occurred between the end of radiation and the start of chemotherapy (8 weeks later). Three tumors (5%) progressed by the end of radiation therapy. Complete disappearance of the enhancing mass was extremely rare, occurring in only three tumors (5%). Response was more common in patients with anaplastic astrocytomas (11 of 21, or 52%) than in those with glioblastomas (11 of 42, or 26%), although this difference did not reach statistical significance.

Because of the poor survival of patients with high-grade astrocytomas, numerous strategies have been tried to improve the results with irradiation. These efforts fall into two main classes: first, increasing radiation dose, and second, modulating the radiation response. Doses higher than 60 Gy have been used, but to no avail. The RTOG 74-01–ECOG 1374 trial randomized between 60 Gy WBRT and a total dose of 70 Gy (60 Gy WBRT followed by a 10-Gy boost) but showed no improved survival with the additional dose (Table 7). Hyperfractionation also has been used to try to increase the total dose. In the Brain Tumor Cooperative Group (BTCG) study 77-02, patients who were randomly assigned hyperfractionated WBRT (66 Gy in 1.1-Gy twice-daily fractions) with BCNU showed no difference in survival compared with those who received conventional WBRT (60 Gy). In RTOG trial 90-06, patients were randomized to receive either 60 Gy in conventional fractionation or 72 Gy in a hyperfractionated regimen (1.2 Gy twice daily). No differences were found in survival between the two groups.

Radioactive implants also have been used to increase dose locally to the tumor bed. Single-institution data suggested improved outcome with this approach over conventional radiotherapy. In the BTCG 87-01 trial, patients were randomized to receive either an 125I seed implant (60 Gy) or no implant at surgery (Table 7). Thereafter, patients received 60 Gy via external beam radiation. Remarkably, no survival benefit was found for use of 120 Gy delivered with brachytherapy and external beam irradiation. In a trial conducted at the Princess Margaret Hospital, patients underwent surgery followed by 50 Gy of external beam radiation and then were randomized to either receive an 125I implant (60 Gy) or not. This study too showed no improvement with an implant.

An implantable balloon has been developed for the delivery of brachytherapy. This implantable balloon (Gliasite) comes in several sizes to fit the diameter of the resection cavity. After resection of the tumor, the balloon is placed in the cavity and filled with x-ray contrast medium. The catheter is attached to the balloon and then is brought out through the skull, and an infusion port at the end of the catheter is attached to the skull. A liquid form of 125I (Iotrex), specifically developed for use in the Gliasite, can then be used to fill the balloon 2 to 3 weeks after insertion. Dwell times range from 2 to 5 days, based on the calculated dose of radiation, following which the Iotrex is withdrawn from the balloon via the infusion port. Loading and unloading of the Iotrex are done percutaneously. Patients can be sent home with appropriate radiation safeguards (personal communication, Allen Sills, MD, Memphis, Tennessee); alternatively, the patient is kept in the hospital with the appropriate radiation isolation precautions. This device provides brachytherapy directly to the resection cavity, with minimal exposure of unaffected brain to the radiation source. The device is used for treatment of primary gliomas as well as metastatic tumors. The dose for recurrent gliomas is 60 Gy at 1 cm and for metastatic lesions 60 Gy at 0.5 cm. Use of the Iotrex has yielded significant improvements in quality of life and tumor control.

Another means of increasing dose to the tumor region after conventional radiation therapy has been stereotactic radiotherapy. Single-institution studies have shown improved results with this approach—for example, from the Joint Center for Radiation Therapy at Harvard. This approach was tested in the RTOG study 93-05, in which patients in the control group received standard fractionated radiotherapy (60 Gy in 30 fractions) with BCNU chemotherapy, whereas patients in the experimental group received a radiosurgery boost (15 to 24 Gy) before external beam radiation therapy. Preliminary results in abstract form from this trial indicate no benefit for the radiosurgery boost (Table 7).

Numerous agents have been tried in combination with radiation in attempts to improve survival. Conventional chemotherapy is discussed later under Chemotherapy for Gliomas. Halogenated pyrimidine analogs are incorporated preferentially into dividing cells, with substitution of pyrimidine for the thymidine in DNA, sensitizing cells to double-strand breaks. In vitro, this can lead to radiosensitization by preventing the repair of double-strand breaks. The halogenated pyrimidine BUdR has been studied in a randomized trial. Prados and associates randomized patients with anaplastic astrocytomas to receive radiation therapy and chemotherapy with procarbazine, CCNU (lomustine), and vincristine (the PCV regimen), with or without the radiosensitizer BUdR, which was given as a 96-hour infusion concurrently with the radiation. No difference in survival was found, however.

Glioblastomas are thought to have regions of hypoxia that can lead to radioresistance. Therefore, agents that are hypoxic cell sensitizers have been used in conjunction with radiation (Table 7). The hypoxic cell sensitizer misonidazole was tested in RTOG trial 79-18 but was not found to lead to an improvement in survival. Misonidazole also showed no benefit in another randomized trial, BTSG 77-02.

The one agent that has clearly been shown to be of benefit when combined with radiation is the alkylating agent temozolomide. The results with this drug for treatment of glioblastomas are discussed next.

Chemotherapy for Gliomas

Chemotherapy for Newly Diagnosed High-Grade Astrocytomas

The role of chemotherapy in the treatment of high-grade astrocytomas has been investigated for more than 30 years. The earliest prospective brain tumor chemotherapy clinical trials were performed by the BTCG/Brain Tumor Study Group (BTSG) investigators. In BTSG 69-01, Walker and colleagues found that median survival was not significantly prolonged in patients receiving carmustine (BCNU), but at 18 months, 19% of patients reciving chemotherapy, radiation therapy, and surgery were alive, compared with 4% of those treated with radiation therapy and surgery alone. In BTSG 75-01, Green showed that the addition of chemotherapy to surgery and radiation therapy significantly increased mean survival from 40 weeks to 50 weeks and increased the percentage of patients surviving 18 months to 24%. In these and most studies of chemotherapy as adjuvant therapy for high-grade astrocytoma, the benefit was greater in patients with anaplastic astrocytoma than in those with glioblastoma.

Another well-studied and frequently used chemotherapy regimen is combination therapy with procarbazine, CCNU, and vincristine (i.e., PCV). Although initial reports suggested a survival advantage for patients with anaplastic astrocytoma receiving PCV over those receiving BCNU, subsequent meta-analysis of four studies failed to confirm this finding. The PCV regimen clearly is more toxic as a result of myelosuppression and peripheral neuropathy and has been replaced by temozolomide.

The introduction of temozolomide has dramatically altered the treatment of high-grade gliomas. Temozolomide is an oral alkylating agent whose levels are not affected by AEDs or other hepatic enzyme-inducing drugs. Toxicity with this drug is relatively mild, with good CNS penetration. The pivotal study conducted jointly by the European Organization for Research and Treatment of Cancer (EORTC) trial 22981–26981, and the National Cancer Institute of Canada (NCIC) trial CE.3 confirmed the usefulness of temozolomide and radiation in newly diagnosed glioblastomas. This study randomized 573 patients to receive 60 Gy of standard fractionated radiation or the same radiation regimen with daily temozolomide at 75 mg/m2 followed by six cycles of adjuvant temozolomide (150 to 200 mg/m2 for 5 days during each 28-day cycle). The patients in the radiation plus temozolomide arm had a statistically significant improvement in survival over those in the radiation only arm (P < 0.001; Fig. 14A ). At follow-up evaluation at a median of 28 months, the median survival periods in the two arms were 12.1 and 14.6 months, respectively. The 2-year overall survival rates were 10.4% and 26.5%, respectively.

Figure 14.  Survival of patients with glioblastoma based on temozolomide administration or MGMT status. Patients enrolled in the EORTC trial 22981–26981 and NCIC trial CE.3 were randomized to receive radiotherapy alone or radiotherapy plus temozolomide. A, Kaplan-Meier estimates of overall survival according to treatment group. The hazard ratio for death among patients who received temozolomide, compared with those who received radiotherapy alone, was 0.63 (95% confidence interval, 0.52 to 0.75l; P < 0.001). B, In a subset of the patients enrolled on the trial in A, the methylation status of the MGMT (O[6]-methylguanine DNA methyltransferase) gene promoter was assessed. The curves in B represent Kaplan-Meier estimates of overall survival according to MGMT promoter methylation status. The difference in survival between patients with a methylated MGMT promoter (92 patients, 65 of whom died) and those with an unmethylated MGMT promoter (114 patients, 105 of whom died) was highly significant (P < 0.001 by the log-rank test), indicating that the MGMT promoter methylation status has prognostic value. In the group of patients with a methylated MGMT promoter, a risk reduction of 55% (hazard ratio for death, 0.45; 95% confidence interval, 0.32 to 0.61) was determined, as compared with the patients with an unmethylated MGMT promoter.  

Hegi and coworkers subsequently evaluated the status of the epigenetic silencing of the MGMT DNA repair gene (MGMT) by promoter methylation in patients enrolled in this trial. Two hundred six of the patients enrolled in EORTC 22981–26981–NCIC CE.3 had samples in which MGMT promoter methylation could be assessed. The MGMT promoter methylation occurred in 45% of these cases, and this was found to be an independent favorable prognostic factor regardless of treatment (P < 0.001). Patients with a methylated MGMT promoter had better survival than those without (see Fig. 14B ). The median and 2-year survival rates for patients receiving radiation versus radiation and temozolomide based on MGMT promoter methylation status are shown in Table 8 . Even among patients with unmethylated MGMT, a trend toward increased survival is recognized for those receiving temozolomide. These data have made irradiation plus adjuvant temozolomide the standard of care for all patients with glioblastoma regardless of MGMT status. The relative contributions of oral daily temozolomide during radiation versus after radiation are not clear, and the optimal dose of temozolomide to inhibit MGMT activity has not been established and is the subject of ongoing trials. In the current RTOG trial, patients with newly diagnosed glioblastoma are randomized to receive standard chemoradiation therapy followed by 12 rather than 6 monthly cycles of adjuvant temozolomide at standard dosing, or, alternatively, dose-intensive temozolomide (21 days on and 7 days off).

Table 8. Results of Treatment for Glioblastomas Stratified by MGMT Promoter Status: EORTC Trial 22981-26981 and NCIC Trial CE.3



Study Result

Unmethylated (N =114)


Methylated (N = 92)

Median overall survival (mo)


(P <0.001)



Radiation only (N = 54)

Chemoradiation (N= 60)

Radiation only (N = 46)

Chemoradiation (N = 46)






2-year survival






(P =0.06)


EORTC, European Organization for Research and Treatment of Cancer; MGMT, O6-methylguanine DNA methyltransferase; NCIC, National Cancer Institute of Canada.

* Either radiation therapy only or chemotherapy plus radiation therapy.

Even though the cited data on temozolomide were generated in patients with glioblastomas, many oncologists have extrapolated these findings to anaplastic astrocytomas and have used this agent concurrently with radiation in this setting. Of note, however, some experts have questioned whether concomitant temozolomide should be considered standard therapy in patients with anaplastic gliomas in the absence of randomized data.

Chemotherapy for Recurrent High-Grade Astrocytomas

Patients with high-grade astrocytomas in whom initial therapy with radiation and temozolomide fails to provide benefit are candidates for participation in phase I/II trials. Such trials include drugs as well as a radiolabeled antibody that is infused directly into the tumor. In patients for whom some interval has elapsed between initial temozolomide therapy and failure, retreatment with low-dose daily temozolomide may be tried. Any benefit from chemotherapy for recurrent disease must be measured against the roughly 4- to 5-month median survival after repeat surgical removal of recurrent high-grade astrocytic tumors.

Some patients with recurrent high-grade astrocytomas are offered re-resection with use of either Gliasite (covered earlier under Radiation Therapy for High-Grade Gliomas) or Gliadel (intracavitary biodegradable BCNU wafers). A placebo-controlled multicenter study in 22 patients reported a median survival period of 31 weeks in patients receiving BCNU polymers, compared with 23 weeks in the patients receiving placebo polymers. The use of Gliadel is limited to those patients who can have meaningful re-resection of tumors that are relatively well circumscribed, unilateral, and in noneloquent brain locations—a situation not typical for a majority of anaplastic astrocytic tumors.

Gliadel wafers also have been used in patients with newly diagnosed high-grade gliomas. In one large study, 240 patients received either BNCU or placebo wafers at the time of primary surgical resection and then underwent external beam irradiation of their tumor. Median survival in the BCNU wafer group was 13.9 months, versus 11.6 months in the placebo group. This modest survival advantage may have been partly outweighed by a higher incidence of adverse effects in the BCNU group, including CSF leak and significant vasogenic edema. Several institutions have reported a high rate of perioperative craniotomy infection after wafer placement.

Irinotecan, a topoisomerase I inhibitor, has been the subject of several studies in patients with recurrent malignant astrocytoma. A respectable 10% to 15% objective imaging response rate has been achieved after thrice-weekly infusions. In two other studies seeking maximal tolerated dose, a marked effect of concomitant enzyme-inducing AEDs was found. Patients taking cytochrome P-450-inducing AEDs achieved lower levels of the active metabolite SN-38 (7-ethyl-10-hydroxycamptothecin) than those measured in patients not receiving AEDs. Subsequent and ongoing irinotecan studies have stratified patients according to AED use.

Chemotherapy for Low-Grade Astrocytomas

Chemotherapy for low-grade astrocytomas usually is reserved for those patients whose unresectable tumors are progressively symptomatic. Radiation therapy traditionally has been used as a first-line treatment for progression. Potential toxicity of any proposed therapy for low-grade gliomas must be balanced against the long natural history. Particularly in younger patients and in those whose tumors have an oligodendroglial component, median survival periods may approach 10 years, with 5-year survival rates of 75% or higher. Indeed, some investigators have argued that quality of life considerations, with particular attention to cognitive status, dictate a prolonged “wait and see” policy in patients with suspected low-grade glioma. Some variants of low-grade glial tumors cary a more ominous prognosis and for the rare situation of multicentric gliomatosis cerebri, chemotherapy may be a reasonable first-line approach. A recent case report documents both MRI and MRS improvement in this usually untreatable condition. Potential considerations in the use of alkylating agents for patients with low-grade astrocytomas should include the risks of hematologic malignancies (5% at 5 years) and sterility. When the astrocytoma contains a portion of oligodendroglial cells, an increasingly common practice is to recommend chemotherapy, but the percentage of oligodendroglial cells required to convey chemosensitivity to the entire mass is unknown. Molecular genetic analysis of oligoastrocytomas has suggested an inverse correlation between 1p or 19q deletions and p53 mutations, rarely occurring in the same tumor. Tumors with 1p or 19q deletions tend to have a predominant oligodendroglial component. It remains to be seen if patients with low-grade tumors can be reliably selected for chemotherapy on the basis of these criteria.

Examination of quality of life outcomes must factor into recommendations for chemotherapy and radiation therapy. The trend over the last 30 years has been for increasing survival times, with a majority of patients receiving only surgery as a first course of treatment. Patients who have a low risk of progression—those who are younger than 40 years of age with tumors less than 4 cm in diameter and who have undergone gross total resection—have an overall 5-year survival rate of 94%, with a progression-free 5-year survival rate of 50%. For these patients, close observation seems indicated.

The Radiation Therapy Oncology Group (RTOG) recently completed trial 9802 for histologically confirmed low-grade gliomas (Fig. 15). Patients who had undergone total resection and were younger than 40 years of age (low-risk group) were managed by observation. The remaining patients were randomized to receive either radiation therapy alone (54 Gy) or radiation therapy followed by six cycles of PCV chemotherapy. The results have been reported in abstract form. For the low-risk patients who were observed, overall survival (OS) at 2- and 5-years was 99% and 94%, respectively, with a median follow-up time of 4 years. Progression-free survival (PFS) at 2- and 5-years was 82% and 50%, respectively. For the high-risk patients who were randomized to RT versus RT + PCV, there was no difference in OS or PFS between the two groups. OS at 2- and 5-years was 87% and 61% (RT alone) versus 86% and 70% with RT + PCV (P = 0.72).”

Figure 15.  Schema for RTOG/Intergroup 9802 protocol for treatment of low-grade gliomas. Protocol was open to patients with histologically verified WHO grade II astrocytoma, oligodendroglioma, or oligoastrocytoma. The study was closed to accrual in June 2002. Results have been reported only in abstract form. CCNU, vincristine; PCV, procarbazine; RT, radiation therapy.

The RTOG currently has a trial (3302) open that is a phase II study for patients with high-risk low-grade gliomas that uses daily temozolomide for 12 cycles.

Chemotherapy for Newly Diagnosed Anaplastic Oligodendrogliomas

Over the past decade, increasing evidence has accumulated to confirm that anaplastic oligodendrogliomas and possibly anaplastic oligoastrocytomas are special types of malignant gliomas that are sensitive to chemotherapy. Although the molecular markers of chemosensitivity, allelic loss of 1p and 19q, have been shown to have prognostic significance in multiple studies, the reason for oligodendroglial chemosensitivity remains unknown. Zlatescu and colleagues have made the clinically interesting observation that allelic loss of 1p and 19q is related to tumor location and extent of tumor spread in the brain. Anaplastic oligodendrogliomas located in the frontal, parietal, and occipital lobes were significantly more likely to harbor 1p or 19q loss than are histologically identical tumors arising in the temporal lobe, insula, or diencephalon. Frontal tumors tend to have greater bilateral diffuse spread, and investigation of the biologic significance of these growth and invasion differences eventually may have implications for management strategies.

In the first large study of chemotherapy for anaplastic oligodendrogliomas, Cairncross and the National Cancer Institute of Canada conducted a multicenter phase II trial of intensive PCV chemotherapy (i.e., procarbazine, CCNU, and vincristine) using higher doses of the first two drugs than are found in the standard PCV regimen. This study included new or recurrent anaplastic oligodendrogliomas and reported a 75% overall response rate, with 38% complete responses. Time to tumor progression was at least 16.3 months for the entire eligible group and at least 25.2 months for complete responders.

Two randomized trials have been conducted comparing radiation only with radiation plus PCV chemotherapy in patients with newly diagnosed anaplastic oligodendrogliomas or anaplastic oligoastrocytomas. In RTOG 9402, patients were randomized to receive radiation only or up to four cycles of intensive PCV chemotherapy followed by radiation. In EORTC 226951, patients were randomized to receive radiation only or radiation followed by six cycles of standard PCV chemotherapy. The two studies showed remarkably similar results. In both studies, the addition of PCV chemotherapy improved progression-free survival but not overall survival. Patients with combined 1p and 19q loss showed a much better survival than that observed for patients with intact 1p and 19q; however, even in this chemotherapy-sensitive subgroup, the addition of PCV did not improve overall survival.

The PCV regimen has significant hematologic toxicity, and given its lack of survival benefit in anaplastic oligodendrogliomas in these randomized phase III trials, there has been a movement toward temozolomide, which is much better tolerated. In spite of a lack of randomized data with temozolomide in anaplastic oligodendrogliomas, many oncologists are currently using this drug concomitantly with radiation, extrapolating from the experience with glioblastoma.

Chemotherapy for Recurrent Anaplastic Oligodendroglioma and Oligoastrocytoma

Most patients with recurrent AOs who received PCV at initial presentation are currently being treated with temozolomide at relapse. Several small trials have shown the safety and efficacy of temozolomide following prior PCV therapy (Table 9). An open label phase II trial of temozolomide in 47 patients with anaplastic oligodendroglioma or anaplastic oligoastrocytoma who relapsed after radiation and PCV chemotherapy demonstrated an objective response rate of 43%, median PFS of 7.5 months and 34% rate of disease-free survival at 12 months. In another phase II multicenter trial that included patients with anaplastic oligodendroglioma or anaplastic oligoastrocytoma, a response rate was seen in 26% of patients who had previously received PCV therapy and in two out of three chemotherapy-naive patients. A short time to tumor progression appears to be a bad prognostic factor for response to both first-line PCV chemotherapy and temozolomide.

Table 9. Results of Treatment for Recurrent Low-Grade and Anaplastic Oligodendroglioma and Oligoastrocytoma: Temozolomide-Based Chemotherapy Regimens


Tumors Treated

No. of Patients


MTP (mo)

6/12-Month PFS


Yung et al














van den Bent et al







Brandes et al







van den Bent et al



53% (CR + PR)




van den Bent et al



25% (CR + PR)

8.0 (in responders)



* This study also included AA and is included here for comparison. Patients with AOA and AO had better survival than those with AA.

In these reports, none of the patients had received any prior chemotherapy before being placed on temozolomide at relapse.

AA, anaplastic astrocytoma; AO, anaplastic oligodendroglioma; AOA, anaplastic oligoastrocytoma; CR, complete response; MTP, median time to progression; MST, median survival time; NR, no response; O, oligodendroglioma; OA, oligoastrocytoma; PFS, progression-free survival; PR, partial response; SD, stable disease.

Brandes and associates recently reported on 67 patients with recurrent anaplastic oligodendroglioma or anaplastic oligoastrocytoma treated with temozolomide at the time of disease progression. All patients had received surgery and radiotherapy, and none had received prior chemotherapy. The overall response rate was 46% and was higher in those with anaplastic oligodendroglioma than in those with anaplastic oligoastrocytoma (62% versus 25%; P = 0.003). Combined 1p and 19q loss (present in 48% of patients) significantly correlated with response rate (P = 0.04), time to progression (P = 0.003), and overall survival (P = 0.0001). By contrast, only a borderline correlation was found between MGMT promoter methylation (present in 69% of assessable patients) and overall survival (P = 0.09).

Chemotherapy for Low-Grade Oligodendrogliomas and Oligoastrocytomas

Temozolomide has been used in patients with low-grade oligodendrogliomas or oligoastrocytomas that have recurred. The EORTC conducted one study using temozolomide for patients who had received PCV chemotherapy and irradiation for their initial disease, and another study for patients who had received surgery and irradiation without chemotherapy initially. These studies showed 25% to 53% response rates, respectively, to temozolomide (Table 9). Loss of 1p has been correlated with response to temozolomide for first-line chemotherapy in patients with recurrent oligodendrogliomas.

Therapy for Elderly Patients with Malignant Gliomas

Elderly patients with malignant gliomas have a poorer outcome than that for younger patients. Brandes and coworkers evaluated the role of surgery, radiation therapy, and chemotherapy for the treatment of newly diagnosed glioblastoma in patients older than 65 years of age. These investigators found that the main predictive factor in evaluating this treatment was perioperative KPS score. The higher the KPS score, the better the patient fared, regardless of the treatment. Patients were stratified into group A (surgery with radiation therapy to a dose of 59.4 Gy), group B (surgery, radiation therapy, and PCV chemotherapy), and group C (surgery, radiation therapy, and temozolomide). The median time to disease progression was significantly better in group C patients. Overall survival was better in group C, but this did not reach statistical significance. Hematologic toxicity was higher in patients on a PCV regimen than in those receiving temozolomide.

Another study evaluated the role of debulking craniotomy versus stereotactic biopsy in the treatment of malignant gliomas in patients older than 65 years of age. Patients in this study were randomized to undergo either stereotactic biopsy or craniotomy for debulking of the tumor. Patients were evaluated with an intention to treat analysis. The median survival period was 171 days after craniotomy versus 85 days after biopsy. The difference was not statistically significant; however, this finding did suggest that patients undergoing a more aggressive craniotomy survived longer. Time to deterioration was 105 days in the craniotomy group and 72 days in the biopsy group. This increased time to deterioration corresponds with maintenance of quality of life during this time period. Both groups of patients tolerated the radiation therapy well.

On the basis of these results, it appears that patients with malignant gliomas older than 65 may still benefit from aggressive therapy if they have a reasonable performance status. Therefore, old age alone should not be a criterion for minimizing therapy. An abbreviated radiotherapy regimen may be considered in older patients with a poor performance status. Roa and associates randomized patients with glioblastoma older than the age of 60 to receive a standard course of radiation (60 Gy in 30 fractions over 6 weeks) or a shorter course (40 Gy in 15 fractions over 3 weeks). Overall survival times measured from randomization were similar at 5.1 months for standard RT versus 5.6 months for the shorter course (P = 0.57). The survival probabilities at 6 months were also similar, at 44.7% for standard RT versus 41.7% for the shorter course.

Quality of Life after Therapy for Gliomas

Quality of life issues are paramount in treatment for patients with gliomas. Taphoorn and coworkers found that drowsiness, fatigue, memory deficits, and concentration problems were more common complaints in patients with low-grade gliomas than in control patients with hematologic malignancies. An analysis of quality of life was performed in patients participating in an EORTC trial of randomization to either 59.4 or 45 Gy for low-grade gliomas. This analysis found that patients receiving the higher dose complained of more fatigue and insomnia immediately after radiotherapy and poorer emotional functioning 7 to 15 months after institution of therapy.

The EORTC/NCIC performed a quality of life study as part of a randomized trial of radiation versus radiation plus temozolomide for patients with glioblastoma. The study investigators found that health-related quality of life measures (fatigue, overall health, social function, emotional function, future uncertainty, insomnia, and communication deficit) did not deteriorate by a clinically meaningful amount in either treatment group over time after treatment, and even improved for some parameters. Furthermore, the addition of the temozolomide regimen to radiotherapy did not negatively affect health-related quality of life. Notwithstanding these results, patients with malignant gliomas face enormous stresses. They face the same burdens of dealing with a terminal disease as patients with breast cancer, lung cancer, and other malignancies, but they have unique perspectives in that they are uniformly concerned about the potential for damage to the brain and the effect of cognitive function. These patients not only have to deal with concepts of dying and leaving behind their families and friends but also have a constant worry about being in a persistent vegetative state and being a burden to their families. Measurement of quality of life is important to elucidate the impact of available treatments on the disease: The goal is to improve survival, but not at the sacrifice of quality of life.

New Approaches to Therapy of Gliomas

The disappointing results with conventional cytotoxic chemotherapy have led to efforts to find more effective and better-tolerated therapies. The challenge of neuro-oncology is to develop new compounds or procedures of gene transfer that specifically target the molecular alterations of glioma tumor cells and restore normal gene expression, cell cycle regulation, and apoptosis. Significant progress is likely to come from some combination of cytotoxic chemotherapy with biologic agents directed at genetic alterations specific to brain tumors, such as growth factor receptors. Receptors are tempting targets because they are extracellular and easily accessible to drugs or antibodies. A number of agents targeted to the EGFR receptor and the vIII variant are being tested. These EGFR inhibitors have shown limited utility in the clinic with brain tumors. These inhibitors may be effective only in patients with tumors that have a susceptible target. In one study, patients who had received treatment with gefitinib or erlotinib were analyzed for expression of EGFR, the deletion mutant EGFRvIII, and the tumor suppressor gene PTEN. Tumors that coexpressed EGFRvIII and PTEN were most likely to respond to EGFR kinase inhibitors.

Receptor tyrosine kinases (RTKs), such as PDGF and insulin-like growth factor (IGFR) receptors, regulate cell proliferation and differentiation. Deregulated RTK signaling frequently is found in human astrocytic tumors. In addition, gefitinib and erlotinib, mentioned earlier, and other inhibitors of RTKs have been evaluated in malignant glioma including imatinib mesylate. Other agents target farnesyl transferase, histone deacetylase, rapamycin (mTOR), protein kinase C, and the ubiquitin-proteosome pathway. In general, small early trials have been disappointing, although the agents were well tolerated. Most are metabolized through the cytochrome P-450–CYP3A4 system, and significant dose modifications are necessary in patients with brain tumors who are taking enzyme-inducing AEDs.

Because gliomas produce specific angiogenic peptides such as VEGF (see the molecular genetics section earlier in this chapter) to stimulate new blood vessel formation, angiogenesis inhibitors are a logical drug development target. Hypoxic areas within gliomas may stimulate angiogenesis and promote invasiveness of glial cells. Many of these changes in gene expression are mediated by hypoxia-inducible factor-1 (HIF-1). Many antiangiogenic agents directed against VEGF, VEGFR, and other targets are in clinical trials. Bevacizumab, a humanized monoclonal antibody against VEGF, showed a 63% radiologic response rate in patients with recurrent high-grade gliomas.

The transfer of genetic material to tumor cells to make them more susceptible to chemotherapy, or to reverse the alterations that sustain the neoplastic phenotype, has been studied in several centers. Gene therapy strategies include transfecting antioncogenes or drug-activating enzymes to glioma cells. The most extensively studied system is the “suicide gene therapy” with herpes simplex thymidine-kinase (HSV-tk) gene inserted into a replication defective murine retrovirus. Murine fibroblasts are engineered to produce these recombinant retroviruses, which are injected into gliomas, infecting the proliferating tumor cells and making them produce thymidine kinase. The tumor cells then are susceptible to the antiherpes drug ganciclovir, while nondividing brain cells are not infected. Clinical studies using this promising approach have been disappointing because of poor tumor cell transfection efficiency, although some long times to tumor progression have been reported. A prospective controlled trial in patients with newly diagnosed glioblastoma showed no improvement in survival or time to tumor progression when intraoperative HSV-tk therapy and standard radiation therapy were compared with surgery and irradiation alone. Adenoviral and adeno-associated viral vector genes are being studied as well. A phase I study of stereotactic injection of an adenovirus vector to transfer wild-type p53 gene is among the early clinical trials now under way that involve viral vectors to replace tumor suppressor genes.

Other translational strategies that have moved to small preliminary human trials include glutamine depletion with phenylacetate and 13-cis-retinoic acid (CRA), a metabolite of beta-carotene that acts to promote differentiation. Yung and colleagues administered CRA orally, but although toxicity was acceptable, the results were not better than with temozolomide as a single agent.

Another strategy involves drugs that inhibit autophagy, such as hydroxychloroquine, which in a single-institution phase II trial resulted in more than doubling of overall median survival. A phase II trial in patients with newly diagnosed glioblastoma will incorporate chemoradiation therapy with temozolomide and adjuvant temozolomide in addition to chronic daily hydroxychloroquine.

Biologic response modifiers that stimulate or restore the immune system include monoclonal antibodies, interferons, and interleukins. Inhibitors of TGF-b2, a cytokine that promotes invasion and angiogenesis, have been shown to inhibit glioma growth in vitro.

Despite effective neuroimaging and earlier diagnosis, advances in the understanding of cellular events that underlie progression of brain tumors, a steady stream of novel agents, and suggestive preclinical animal studies, clinical trials have proved disappointing. Because of the heterogeneity of malignant glial tumors, treatment strategies probably will require synergistic combinations of cytotoxic agents and noncytotoxic specific molecular methods, with the hope that genetic profiling eventually will help pinpoint appropriate choices for individual patients.


Histopathologic Features

Primary CNS lymphoma (PCNSL) has been known previously by many other names, including “reticulum cell sarcoma,” “diffuse histiocytic lymphoma,” and “microglioma.” The cell of origin is the B lymphocyte, and PCNSL is therefore a high-grade non-Hodgkin's B-cell neoplasm, usually diffuse large cell or large cell immunoblastic type. A frequent finding is perivascular clusters of lymphocytes, and T-lymphocyte infiltrates are common in immunocompetent patients. Although the Epstein-Barr virus genome has been found in HIV-infected patients with PCNSL, no consistent confirmation of viral etiology of PCNSL has been possible among immunocompetent patients. Human herpesvirus 8 (Kaposi's sarcoma-associated herpesvirus) was reported in over 50% of both normal and immunocompromised patients with PCNSL. Human T-cell lymphotropic virus type 1 (HTLV-1) and hepatitis C virus have not been associated with PCNSL.

Only 1% to 3% of PCNSLs are of T-cell origin. The incidence of T-cell PCNSL appears to be higher in Japan than in the United States. Primary T-cell lymphoma has been diagnosed in both immunocompromised and immunocompetent patients. A younger age at diagnosis for T-cell lymphomas has been suggested, and presentation as an infratentorial lesion appers to be more frequent, with one third to one half of reported cases in the cerebellum or brainstem. T-cell lymphoma may appear as a large cystic mass on MRI. The prognosis for T-cell PCNSL may be worse than that for comparably staged B-cell tumors. Authors of published case reports suggest that this tumor should be treated aggressively with regimens used for the more common B-cell lymphomas.

Tumor Biology

PCNSL arises in the brain leptomeninges, spinal cord, or eyes and rarely spreads outside the central nervous system. Its predilection for the periventricular white matter gives rise to the characteristic neuroimaging appearance of a hyperdense mass on unenhanced CT or a hypointense appearance on long TR-weighted MRI. Three fourths of immunocompetent patients have a solitary enhancing mass lesion at presentation. After contrast administration, most bulky masses of PCNSL enhance, most often homogeneously but occasionally in a ring-like pattern (Fig. 16). PCNSLs, however, appear to be diffusely infiltrative at the time of presentation. These areas of disease are not visible on neuroimaging studies because they are behind a relatively intact blood-brain barrier. Postmortem correlates with findings on MRI performed shortly before death show widespread microscopic infiltration in areas that appear normal on the MRI scan. Therefore, PCNSL can be classified as stage 1E disease and should be considered for treatment purposes to be a whole-brain disease.

Figure 16.  Central nervous system lymphoma seen on magnetic resonance imaging. A, T1-weighted post-gadolinium axial image shows a homogeneously enhancing frontal mass compressing the ventricles. B, Corresponding fluid-attenuated inversion recovery (FLAIR) image shows extensive peritumoral edema.

Until the 1980s, PCNSL was considered a rare tumor, accounting for approximately 2% of CNS malignancies in immunocompetent patients and 1% to 2% of all lymphomas. The demographics of the disease have changed, however, both among immunocompetent patients and among those with HIV infection and organ transplant recipients. The incidence of PCNSL among immunocompetent persons has increased 25-fold to 51 cases per 10 million by 2000. Among immunocompetent patients with PCNSL, median age at diagnosis is 55 years, and males outnumber females by 2 : 1. Among patients with AIDS, the median age is 35 years, and 95% of HIV-infected patients with PCNSL are male. Among organ transplant recipients, the peak incidence of PCNSL is at 6 months after transplantation, a period much shorter than in the pre-cyclosporine transplantation era.

Recent data from the Surveillance, Epidemiology, and End Results (SEER) program show an overall decline in total PCNSL incidence rates, from a peak of 102 per million in 1995 to 51 in 1998, a decrease largely attributable to decline in the disease in males younger than 59 years of age. The annual rate among patients older than 60 has remained unchanged since 1994. The decline in the younger male population with PCNSL reflects the advent of effective antiretroviral therapy and a declining incidence of PCNSL in the HIV-infected population. This chapter discusses PCNSL and its treatment in the immunocompetent population.

Clinical Diagnosis and Staging

The most common clinical presentation is one of progressive focal symptoms. Seizures also may occur. Less commonly, progressive cognitive decline without focal symptoms leads to the diagnosis. Several variant clinical presentations are possible in the immunocompetent population: primary ocular, meningeal, or relapsing-remitting disease; intravascular malignant lymphomatosis with stroke-like onset; or neurolymphomatosis with both peripheral and CNS involvement. Considerations in the differential diagnosis for PCNSL include high-grade glial tumor, CNS metastases, neurosarcoidosis, and tumefactive multiple sclerosis. Spinal cord PCNSL is very rare. Ocular lymphoma may be the first manifestation of the disease or its relapse. Patients with ocular lymphoma have a 50% to 80% chance of developing cerebral lymphoma.

Initial evaluation of patients with suspected PCNSL should include a thorough physical examination to exclude possible extraneural sources of lymphoma. Neurologic evaluation is directed at clarification of the extent of CNS disease. Extraneural sites are uncommon in patients without prior known lymphoma, and extensive abdominal and pelvic studies usually are not warranted. Chest x-ray studies should be performed, and blood tests should include a complete blood count (CBC) and liver function tests, as well as HIV testing. Gadolinium-enhanced spinal MRI should be performed, because bulky nodular disease sometimes can be seen in the leptomeninges.

Ophthalmologic consultation for slit-lamp examination is a critical part of the workup, because up to 10% to 15% of patients with PCNSL will have vitreous involvement at the time of diagnosis, and half of these will have no visual symptoms. Lumbar puncture should be performed if not contraindicated because of the intracranial mass location(s). Glucose and protein determinations, cell count, and flow cytometry should be performed. Recent reports of molecular diagnosis of PCNSL by the demonstration of monoclonality by amplification of the rearranged IgH genes by polymerase chain reaction (PCR) to the CDR-III region raises the possibility of definitive diagnosis even with few cells present in the CSF. In several cases, however, the PCR analysis was negative even when conventional cytologic findings were suggestive of malignancy. In another study, monoclonal patterns in the CSF were found in 77% of biopsy-proven cases, with no false positives among control subjects. At present, however, it is premature to base clinical therapeutic decisions on these promising early findings.

When the foregoing procedures fail to confirm the diagnosis, the surgical procedure of choice is a stereotactic brain biopsy. Aggressive surgical resection does not improve survival and may cause deterioration owing to the deep location of many PCNSLs. It is most important to try to withhold corticosteroids from patients during initial evaluation and surgical confirmation. Corticosteroids repair the blood-brain barrier and also have a direct cytolytic effect on B-cell lymphomas. Although the patient's clinical symptoms may abate with early institution of corticosteroids, with frequent nearly complete resolution of MRI-enhancing abnormalities, diagnosis will be compromised, and corticosteroids will have to be withdrawn in order to proceed with definitive biopsy confirmation. Recent interest has focused on immunohistochemical definition of morphologic markers of prognostic significance on the biopsy material (Fig. 17). Braaten and colleagues studied the expression of BCL-6 antigen in PCNSL and found that the presence of this antigen predicted a median survival of 101 months, compared with 14.7 months in patients without this marker. Other groups of investigators, however, have found that the marker correlates with a negative prognosis, so clinical decisions cannot be made on the basis of early immunohistochemistry studies.

Figure 17.  Central nervous system lymphoma: histologic features. A, High-grade B-cell lymphoma with dense perivascular lymphocytic cuffing. (Hematoxylin and eosin staining.) B, Dense staining with CD20 (a B-cell marker) in the same specimen as in A.


PCNSL is an aggressive disease for which median survival in immunocompetent patients is only 3 months without treatment. Increasing age and poor performance status are important negative prognostic variables. The goal of treatment is to eradicate both contrast-enhancing mass lesions and microscopic infiltration of brain, spine, leptomeninges, and vitreous. Treatment must be designed to maximize efficacy but also to minimize toxicity to the brain. Optimal treatment has not been established. The reason for the lack of standardized treatment protocols is the relatively small number of patients and the absence of a phase III randomized trial. In the past 15 years, however, steady increases in median survival have been achieved in patients with PCNSL. Five-year survival rates on the order of 40% are now being reported by many groups of investigators, with median survival periods of 3 to 4 years.

Radiation treatment alone led to median survival periods of 18 months in early clinical trials, with only 3% to 4% 5-year survival rates. Chemotherapy was first used as an adjunct to radiation therapy more than 20 years ago after methotrexate was found to be effective in the treatment of systemic lymphoma. Twelve studies totaling more than 450 patients have established the efficacy of multiple chemotherapeutic regimens. Initially, most were used along with or after a course of whole-brain radiation therapy. The most commonly used drugs have been methotrexate, cytarabine, and cyclophosphamide. The first two agents have been administered intravenously and intrathecally. Extension of median survival periods for up to 4 years with combined regimens was achieved in the mid-1990s. The most frequently used combined regimen involved administration of intravenous methotrexate at doses ranging from 3.5 to 8 g/m2 every 10 to 14 days for three cycles, followed by whole-brain radiation therapy to a dose of 40 to 50 Gy, with subsequent administration of cytarabine 3 g/m2 for three cycles. DeAngelis and colleagues reported a 58% complete response rate, with additional 36% rate of partial responses and progression-free median survival for 24 months. A worse prognosis was associated with age older than 60 years (survival for 50.4 months versus 21.8 months). With improved survival, however, severe delayed neural toxicity has developed in increasing numbers of patients. At 18 months, 15% to 50% of patients who had received combined radiation therapy and chemotherapy were found to have extensive white matter abnormalities, with consequent severe cognitive decline. Age and pre-existing white matter disease due to hypertension may be relative risk factors. This complication typically begins from 4 months to several years after treatment. Recognition of the serious sequelae of combined radiation therapy and chemotherapy led to attempts to provide chemotherapy as the sole modality for treatment of newly diagnosed PCNSL. Initial response rates to chemotherapy vary from 50% to 100%, with duration of response between 12 and 44 months. Typical of the successes of many clinical trials are the results of Cher and colleagues, who reported complete responses in 17 of 19 patients receiving methotrexate alone, with an event-free median survival period of 32 months and overall survival period of 53 months. Therapeutic levels of methotrexate can be achieved in the CSF after intravenous drug administration, making it possible to achieve clearance of malignant cells from the CSF without intrathecal administration.

Attempts to improve on the results of methotrexate have included a variety of other multidrug chemotherapy regimens. Standard regimens effective in the treatment of comparable systemic non-Hodgkin's lymphomas (CHOP, CHOD, or MACOP-B) are not effective for PCNSL. Attempts to disrupt the blood-brain barrier with hyperosmolar intra-arterial mannitol do not provide additional survival benefit beyond that achieved with intravenous methotrexate-based regimens.

Many neuro-oncologists, therefore, have concluded that high-dose methotrexate should be offered to all patients as the first-line agent for treatment of PCNSL. However, there are dissenting opinions. Herrlinger and colleagues recently reported only a 29.7% complete response rate with intravenous methotrexate while 37.8% of their patients progressed, and they revisit the question of whole brain radiation therapy as a first-line treatment for PCNSL. Most centers, however, have based their therapy on methotrexate regimens. Long-term follow-up of patients who achieved durable remissions for more than 1 year suggests that chemotherapy alone is associated with a low risk of neurotoxicty even in the elderly, although MRI may show significant areas of clinically asymptomatic leukoencephalopathy. The neuropsychological outcome after chemotherapy alone for primary CNS lymphoma is quite good, whereas increasing numbers of studies document progressive dementia, sometimes with gadolinium-enhancing lesions on MRI, as early as several months after combined-modality treatment. Autopsy studies have confirmed not only widespread “pan-brain” infiltrative lymphoma but extensive vascular changes and leukoencephalopathy in patients who received radiation with or without chemotherapy.

The following recommendations apply to immunocompetent patients only. The goal of treatment is to achieve a complete response while avoiding irradiation of normal brain. Patients with AIDS-related PCNSL require individualized treatment based on their immune status and presence of concurrent infections. Transplant recipients similarly require an initial attempt at reduction of immunosuppression, and their treatment also must be tailored to the extent of organ dysfunction due to the primary disease process and concurrent therapy.

No firmly established dose for methotrexate has been confirmed; doses in excess of 3.5 g/m2 have been shown to achieve satisfactory CSF drug levels. Figure 18 presents an algorithm for the diagnosis and management of PCNSL at the Hospital of the University of Pennsylvania; the recommended approach to management of these tumors is summarized in Box 2; and Figure 19 presents an algorithm for treatment of these tumors. Patients with PCNSL receive methotrexate 8 g/m2 every 14 days for up to 8 cycles. The calculated dose is diluted in 500 mL of 5% dextrose in water (D5W) and given intravenously over a 4-hour period. Patients achieving a complete response will receive 2 additional doses of methotrexate at 14-day intervals, followed by 11 monthly doses of methotrexate 8 g/m2 so long as a complete response is maintained. MRI scans are obtained monthly during induction therapy and at 3-month intervals during maintenance therapy. The methotrexate protocol involves close clinical monitoring, adjustment of intravenous fluid, and calcium leucovorin rescue, in addition to frequent monitoring of urine pH, renal function, and methotrexate levels.

Figure 18.  Algorithm for diagnosis and management of primary central nervous system lymphoma (PCNSL) at the University of Pennsylvania. CBC, complete blood count; CSF, cerebrospinal fluid; CT, computed tomography; HIV, human immunodeficiency virus; MRI, magnetic resonance imaging; SLE, slit-lamp examination. *Enhanced scan demonstrating periventricular lesions with homogeneous enhancement.

Box 2.


  • Primary central nervous system lymphoma (PCNSL) is a non-Hodgkin's B cell lymphoma arising in the brain, leptomeninges, spinal cord, or eyes. It should be regarded as a whole-brain disease; it rarely spreads outside the nervous system.
  • Optimal treatment of PCNSL is not yet fully established, but experience acquired during the past decade with various combinations of radiation therapy and chemotherapy suggests that initial induction treatment should be attempted with a methotrexate-based intravenous chemotherapy regimen and that concurrent intrathecal chemotherapy is not necessary. Median survival times longer than 40 months have been reported with methotrexate monotherapy, with significantly less cerebral white matter toxicity and its clinical correlate of cognitive decline, than seen in earlier protocols that included cranial irradiation.
  • Intravenous methotrexate 8 g/m2 every 14 days for up to 8 cycles is used for induction therapy, followed by 11 monthly cycles of methotrexate at the same dose in patients in whom a complete response is attained. Leucovorin rescue also is given, and concurrent corticosteroids and antiepileptic drugs as indicated are prescribed.
  • Adequate delivery of chemotherapy to microscopically infiltrated areas of brain protected by the blood-brain barrier remains a significant challenge, and recurrence at multiple cerebral sites remains a problem for which effective therapy remains to be developed.
  • Retreatment of a prior complete responder at relapse with the foregoing intensive methotrexate regimen is feasible. Alternatives include whole-brain radiation therapy, cytarabine, and investigational regimens using temozolomide or rituximab.

Figure 19.  Algorithm for definitive treatment of primary central nervous system lymphoma. MRI, magnetic resonance imaging.

Contraindications to methotrexate therapy include allergy to methotrexate; inability to achieve adequate hydration because of cerebral edema, or cardiac or pulmonary problems; and concurrent immunosuppressive treatment. Patients should not have received prior cranial irradiation. Patients with renal dysfunction resulting in a creatinine clearance of less than 50 mL/min or serum creatinine level greater than 2 mg/dL should not receive methotrexate. Patients with significant ascites or pleural effusions may experience delayed methotrexate clearance because of third space accumulation.

Calcium leucovorin rescue therapy should begin 24 hours after the start of methotrexate infusion. The dose is adjusted according to methotrexate levels. Any dose of leucovorin greater than 50 mg should be given intravenously. If the plasma concentration of methotrexate at 24 hours is greater than 10-5 M, 100 mg/m2 of leucovorin is given intravenously every 6 hours until rescue is achieved and continued at 25 mg intravenously or orally every 6 hours until the methotrexate plasma concentration is less than 10-7 M. In most patients, the methotrexate concentration clears to this level by 72 hours after the infusion.

Patients with an established PCNSL diagnosis may receive concurrent corticosteroid therapy to alleviate symptoms. Concurrent use of salicylates or other nonsteroidal anti-inflammatory drugs or sulfonamide medications is prohibited for at least 1 week before the initiation of methotrexate therapy. Many patients with PCNSLs will remain on corticosteroids for extended periods. The combination of corticosteroids and methotrexate may lead to a low CD4+ count, and these patients will be at risk for Pneumocystis jiroveci pneumonia. Like patients with other brain tumors, patients with PCNSL should receive prophylaxis with trimethoprim-sulfamethoxazole two or three times per week, with discontinuation of this drug 1 week before institution of methotrexate therapy. Concurrent AED therapy is acceptable, and management of nausea associated with chemotherapy does not differ in this population from that for other patients receiving comparable regimens.

The management of progressive or recurrent PCNSL is not yet well established. Age and performance status of the patient must be taken into account. In general, greater than 25% enlargement of previous areas of gadolinium contrast enhancement, the appearance of new lesions, or the appearance of malignant cells in the CSF, vitreous, or, rarely, elsewhere in the body constitutes treatment failure. Cher and colleagues have reported that for patients who have completed their maintenance therapy, return to a more intensive methotrexate dosage regimen may lead to a second complete response. For patients who have progressed through therapy, other chemotherapeutic agents such as cytarabine given intravenously or intrathecally may be considered. At relapse or in the case of failure to achieve complete response after 8 cycles of methotrexate, many treating physicians would consider palliative radiation therapy. The immediate palliation of progressive signs and symptoms is a realistic goal but must be weighed against the possibility that survival will be extended enough to allow emergence of late cognitive neurotoxicity. At most centers, whole-brain irradiation to 40 to 45 Gy in 20 doses is given over 4 to 5 weeks.

After initial diagnosis, surgical intervention usually has little place in treatment. The need for intrathecal chemotherapy, however, may mandate placement of an Ommaya reservoir. Development of communicating hydrocephalus with or after treatment of meningeal lymphoma may require shunting.

Parenchymal treatment failure is the most common pattern of relapse. Ocular, meningeal, and late rare extra-CNS relapses (breast, abdominal wall, bone, lymph nodes) have been described and may be seen with increasing frequency as patients survive longer. Temozolomide is under investigation as an agent for treatment of progressive PCNSL. Rituximab, a monoclonal antibody against the B cell-specific CD20 antigen, has been demonstrated to be effective against various non-Hodgkin's lymphomas, but CSF drug levels attained with intravenous therapy are not high. Attempts to circumvent this problem have involved intraventricular administration through an Ommaya reservoir, with some early reports indicating total clearing of tumor cells with leptomeningeal lymphoma. Whether rituximab will clear parenchymal masses is not clear. Intensive chemotherapy followed by hematopoietic stem cell rescue using a regimen of cytarabine and etoposide has been reported to give a complete response rate of 70%, with a median progression-free survival period of 3 years. Several of the patients in the reported study, however, had intraocular lymphoma only.


Clinical and Pathologic Considerations

Meningiomas account for approximately 30% of all intracranial tumors (Table 1). The male-to-female ratio is 1 : 2, with the incidence increasing with age and reaching a peak in the seventh decade. These tumors arise from arachnoidal cells in the meninges, not the brain parenchyma; therefore, they are extra-axial. They produce signs and symptoms by compressing normal tissues. The locations of meningiomas in two large series are shown in Table 10 . The specific signs and symptoms depend on the anatomic location of the meningioma. For example, meningiomas arising from the cerebral convexity can cause altered mentation and seizures, whereas those arising from the suprasellar region are likely to cause loss of vision, bitemporal hemianopia, and optic atrophy.

Table 10.  Meningiomas: Distribution by Anatomic Location



Tumor Location

Series 1[*]

Series 2[†]







Sphenoid ridge



Lateral ventricle




Cerebellar convexity/posterior fossa









Cerebellopontine angle


Olfactory groove



Foramen magnum








*  Of 179 cases reported by Rohringer and associates; this series included only cases of intracranial meningiomas, none of the spine.

Of 225 cases reported by Mirimanoff and coworkers.

Numerous potential etiologic agents for meningiomas have been investigated, including radiation, trauma, viruses, occupational exposure, diet, and exposure to sex hormones. Only ionizing radiation, however, has been strongly implicated in the pathogenesis of these tumors—for example, after scalp irradiation for tinea capitis. Some relationship with breast cancer may exist, because the likelihood of developing a meningioma after developing breast cancer, or breast cancer after meningioma, is higher than in the general population.

Deletions of chromosome 22 are frequent in meningiomas. The likely target of this chromosomal deletion is the NF2 gene, which is mutated in patients with neurofibromatosis type 2 (NF-2). One of the tumor types commonly seen in these patients is meningioma. The NF2 gene also appears to have an important role in the pathogenesis of sporadic meningiomas. One study found NF2 mutations in 60% of sporadic meningiomas, all of which had lost one copy of chromosome 22. However, 40% of meningiomas have neither allelic loss of chromosome 22 nor NF2 gene mutation; therefore, a second tumor suppressor gene probably is linked to the development of meningiomas.

Grossly, meningiomas appear as rounded masses, well circumscribed in contour, with a well-defined dural base that can be easily separated from the underlying brain tissue (Fig. 20). Often these tumors extend into the adjacent bone. Histologically, several different subtypes of benign meningiomas are recognized, including syncytial, fibroblastic (fibrous), transitional (mixed), psammomatous, microcystic, and papillary; however, these distinctions carry little prognostic significance. Of more importance prognostically is whether the meningioma is benign, which it is in 90% of the cases. Features of atypical meningiomas are focal isolated necrosis, prominence of nucleoli, and mitotic figures with high cell density. Malignant meningiomas are characterized by invasion into adjacent normal brain tissue. In one large study based on data from the U.S. National Cancer Data Base (NCDB), the overall 5-year survival rates in patients with benign, atypical, and malignant meningiomas were 70%, 75%, and 55% respectively.

Figure 20.  Meningioma: gross specimen. A large-convexity meningioma severely displaces the underlying tissue downward and laterally, creating a midline shift and resulting in marked ventricular compression.  

Meningiomas have a distinctive appearance on radiologic studies. On MRI, most meningiomas are isointense with gray matter on T1-weighted scans and enhance intensely with gadolinium (Fig. 21). Peritumoral edema is evident in 60% of meningiomas, and associated bony changes, either destruction or hyperostosis, are seen in 20%. The bony changes are better visualized on CT scans than with MRI. Within the tumor may be seen calcifications, central necrosis, or pseudocysts.

Figure 21.  Meningioma. Axial T1-weighted post-gadolinium magnetic resonance image shows an extra-axial enhancing mass that compresses the underlying brain tissue.

Surgery and Conventional Radiation Therapy for Meningiomas

In a classic paper describing the outcome in 225 patients with meningiomas treated with surgery as the sole modality at Massachusetts General Hospital, Mirimanoff and associates found progression-free survival rates of 93%, 80%, and 68% at 5, 10, and 15 years, respectively, for those who underwent a total resection versus 63%, 45%, and 9% for those who underwent a subtotal resection (P < 0.0001). In general, lower rates of progression were seen with sites associated with higher resectability rates. For example, meningiomas in the convexity, 96% of which underwent total resection, had a 5-year recurrence or progression rate of 3%, versus 34% for lesions in the sphenoid ridge, which had only a 28% rate of total resection. These results suggested that surgery alone was inadequate therapy for meningiomas if total resection could not be achieved. In a subsequent study from Massachusetts General Hospital, Miralbell and coworkers reported on 17 patients who underwent subtotal resection of a meningioma followed by radiation therapy. Their 8-year progression-free survival rate was 88%, which was much better than the 48% rate calculated for the cohort of patients with subtotal resections who did not receive postoperative radiation.

Another retrospective study from the University of California at San Francisco reached similar conclusions regarding the improvement in local control with postoperative radiation therapy in patients who have undergone a subtotal resection. This series included mostly patients with benign meningiomas. A much poorer outcome was observed for patients with malignant meningiomas than for those with benign meningiomas (5-year progression-free survival rate of 89%, versus 48%; P = 0.001).

The foregoing studies all have used conventional fractionated radiation therapy, usually to doses ranging from 50 to 60 Gy of radiation via conventional fractionation and delivery techniques. Most investigators today would recommend approximately 54 Gy for benign meningiomas after incomplete resection and up to 60 Gy for meningiomas that have atypical or malignant features.

Stereotactic Radiation Techniques for Meningiomas

Substantial experience with use of stereotactic techniques in the treatment of meningiomas has been accumulated. One of the centers with extensive experience with this technique is the University of Pittsburgh: In a long-term follow-up study of 99 patients with meningiomas treated with Gamma Knife stereotactic radiosurgery from 1987 to 1992 with 9 to 25 Gy (median marginal dose of 16 Gy), the total rate of failure was 11% at 63 to 120 months after stereotactic radiosurgery. A majority of meningiomas showed a decrease in size by MRI with time. At years 4 to 6, 69% of cases showed a decrease in size by MRI scan; by years 8 to 10, this number had increased to 88%.

The Mayo Clinic and the Brigham and Women's Hospital also have reported rates of local control on the order of 89% in benign meningiomas with use of Gamma Knife- and linear accelerator-based stereotactic radiosurgery. Follow-up remains short, however, with a median duration of 31 to 40 months. Furthermore, in both of these series, patients with atypical or malignant features did poorly. In the Mayo Clinic series, the 5-year local control rates were 93%, 68%, and zero for benign, atypical, and malignant meningiomas, respectively (P < 0.0001). Despite the goal to deliver an extremely conformal dose, stereotactic radiosurgery can cause significant complications. Rates of complications, often involving cranial nerves, have been reported in the 5% to 13% range. With the reduction in radiation dose used to treat meningiomas, however, this complication rate has decreased.

A large experience has been reported for use of stereotactic radiosurgery for treatment of cavernous sinus meningiomas. Because of their location, surgical resection can cause significant complications, particularly cranial nerve morbidity and resulting extraocular muscle paralysis. A number of institutions have reported high local control rates, in the 91% to 98% range, using marginal doses that typically range from 12 to 18 Gy, but with very short follow-up (median, 2 to 3 years). The current standard dose delivered with stereotac tic radiosurgery or Gamma Knife for treatment of these tumors is a marginal dose of 14 Gy. Furthermore, cranial nerve V appears to be sensitive to single high-dose fractions, because the incidence of trigeminal nerve dysfunction has ranged from 4% to 11% in these series when higher doses are used.

In an effort to decrease late effects, fractionated stereotactic radiotherapy (FSRT) also has been used to treat meningiomas. At the University of Heidelberg, a dose of 56.8 Gy was given in 1.8-Gy daily fractions, delivered using a relocatable headframe, for treatment of large base-of-skull meningiomas. At a median follow-up time of 35 months, the 5-year progression-free survival rate was 94% for 180 patients with benign meningiomas but 78% for those with atypical features. Another instance in which the fractionated technique has been used instead of conventional stereotactic radiosurgery is for treatment of optic nerve sheath meningiomas, because of the potential risk of optic nerve injury with a single large dose of radiation. One group of investigators reported using FSRT with 50 to 54 Gy in 1.8-Gy fractions to treat these tumors. Of 22 optic nerves with vision before FSRT, 20 nerves (92%) demonstrated preserved vision, and 42% manifested improvement in visual acuity and/or visual field at follow-up.

A nonrandomized comparison of conventional (nonstereotactic) fractionated radiotherapy versus Gamma Knife radiosurgery for selected patients with cavernous sinus meningiomas was reported by Metellus and coworkers. Thirty-eight patients received fractionated radiotherapy, and 38 had Gamma Knife radiosurgery. Both groups fared well, with progression-free survival rates of 94.7% and 94.4%, respectively. Permanent morbidity was seen in 2.6% of patients in the first group and in none in the second group. These results suggest that in selected patients, Gamma Knife radiosurgery can be performed with little late morbidity.

In summary, numerous reports have described the use of stereotactic techniques for delivering radiation to meningiomas as an alternative to surgery or after subtotal resection. In view of the long natural history of these tumors, however, much longer follow-up will be needed to adequately evaluate these modalities.

Medical Therapy for Meningiomas

A number of agents, including hydroxyurea, interferon-a, tamoxifen, and mifepristone (RU-486), have been reported to be provide modest benefit in patients with recurrent meningiomas. Isolated reports have indicated reduction in the size of meningiomas with hydroxyurea and RU-486. Other studies have not always reported regression, although in some cases, treatment with either of these agents appears to halt the growth of the tumor or to ameliorate the patient's symptoms, or both. Box 3 summarizes the recommended approach to management of meningiomas.

Box 3. 


  • Meningiomas are extra-axial tumors that arise from dura; common locations are cerebral convexity, parasagittal falx, and sphenoid ridge.
  • A very long natural history is characteristic, mandating prolonged follow-up.
  • Greater than 90% of these tumors are benign; the remainder exhibit atypical histologic features or frank invasion of brain parenchyma.
  • Primary treatment is surgical, if feasible.
  • Radiation therapy is reserved for tumors that are incompletely resected, recur after surgery, are inaccessible to surgical resection, or have atypical or invasive features.
  • The standard radiation dose has been approximately 54 Gy for benign meningiomas and up to 60 Gy for those with atypical or invasive features.
  • Stereotactic radiation therapy techniques have been used; however, follow-up is still short in these studies.
  • Anecdotal reports exist of responses to medical therapy (hydroxyurea and antiestrogen and antiprogesterone agents).


Clinical and Pathologic Considerations

Pituitary adenomas make up 10% to 15% of all intracranial tumors. They arise from the anterior lobe of the pituitary gland. The pituitary gland and stalk normally are isointense with brain on T1-weighted MRI scans. The gland and stalk, however, intensely enhance after contrast administration because of the absence of a blood-brain barrier. Pituitary adenomas generally are seen as hypointense foci on T1-weighted MRI scans and do not enhance with gadolinium. Other abnormalities that can appear as a hypointense, nonenhancing lesion in the pituitary include pars intermedia cysts, metastases, infarctions, and epidermoid cysts and abscesses.

Asymptomatic pituitary adenomas were found in 1.5% to 27% of autopsy cases and in 10% of normal adult volunteers by MRI scanning. Symptomatic pituitary adenomas are much less common. These tumors can give rise to signs and symptoms through secretion of hormones or by compression of nearby structures, causing neurologic disturbances such as headaches, bilateral bitemporal hemianopia from compression of the optic chiasm, and cranial nerve palsies from invasion into the cavernous sinus. Adenomas also can cause hypopituitarism from compression of the pituitary stalk.

Many investigators have studied the genetics of pituitary adenomas. Patients with multiple endocrine neoplasia type 1 (MEN-1) are predisposed to developing pituitary adenomas as well as parathyroid and pancreatic islet tumors. Anterior pituitary tumors occur in 30% of patients with MEN-1, most commonly prolactinomas but also nonfunctional and growth hormone (GH)- and adrenocorticotropic hormone (ACTH)-secreting tumors. MEN1 mutations are not commonly found in sporadic pituitary adenomas; however, one gene that is mutated in a substantial proportion of pituitary adenomas is that encoding the a subunit of the guanosine triphosphate (GTP)-binding protein Gs. Mutation of this gene leads to constitutive activation of the cyclic adenosine monophosphate (cAMP) pathway. The mutated form of this gene is called gsp and is present in 10% to 40% of GH-secreting adenomas.

At one time, pituitary adenomas were classified according to their staining characteristics (basophilic, eosinophilic, chromophobe). Today, however, they are classified according to secretion of hormones. In one series of 684 patients with pituitary adenomas who underwent surgery, prolactinomas were the most common (43%), followed by nonsecreting tumors (30%) (Table 11). Thyroid-stimulating hormone (TSH)-secreting tumors were rare in this series (2 of 684), as they are in others.

Table 11. Pituitary Adenomas: Clinical Presentation by Endocrine Secretion


Frequency[*] (%)

Signs and Symptoms

Typical Size at Diagnosis



Women: Amenorrhea, galactorrhea

Women: Microadenoma



Men: Impotence, hypopituitarism

Males: Macroadenoma

Nonsecreting gonadotropin






Gigantism in children




Acromegaly in adults




Cushing's syndrome




Nelson's disease





Can be microadenoma; often macroadenoma owing to delayed diagnosis

ACTH, adrenocorticotropic hormone; GH, growth hormone; TSH, thyroid-stimulating hormone

*  Among 684 cases reported by Oruckaptan and colleagues.

Adenomas also are classified on the basis of size as either macroadenomas (greater than 1 cm in diameter) or microadenomas. Nonsecreting adenomas generally are macroadenomas at the time of diagnosis because they usually come to medical attention fairly late, when they are causing neurologic symptoms due to mass effect. The same is true for gonadotropin-secreting adenomas, which are inefficient producers and secretors of hormones. GH-secreting adenomas and prolactinomas in men often are macroadenomas at the time of diagnosis, probably because their clinical effects evolve over time and often are ignored early on. Because the same pituitary stem cell can produce both GH and prolactin, many patients have adenomas that secrete both hormones. Prolactinomas in women and ACTH-secreting adenomas usually are microadenomas at diagnosis. Prolactinomas in women commonly cause amenorrhea and galactorrhea. ACTH-secreting adenomas often cause dramatic clinical signs and symptoms associated with hypercortisolism that rapidly bring patients to medical attention. ACTH-secreting pituitary adenomas also can be seen in approximately 25% of patients who have undergone bilateral adrenalectomies for Cushing's syndrome secondary to loss of negative feedback control by cortisol on the hypothalamus. In a majority of these patients, hyperpigmentation develops as a result of ACTH hypersecretion. This is termed Nelson's syndrome, and in this setting the pituitary tumors usually are very large and often are difficult to completely resect.

Surgery for Pituitary Adenomas

The current therapy for most pituitary adenomas, excluding prolactinomas, involves surgery. For nonsecreting tumors, which tend to be large and manifest with neurologic signs and symptoms, surgery offers rapid decompression of the visual pathways. For hormonally active tumors, surgery leads to a rapid drop in hormone secretion. The current preferred technique is the transsphenoidal approach, which was popularized by Harvey Cushing in the early 1900s. It then fell out of favor but regained popularity in the 1960s. Transsphenoidal resection is fairly safe, with a mortality rate less than 1%. The most common complications arising from this surgery are nasal septum perforation, anterior pituitary insufficiency, postoperative diabetes insipidus (which usually is transient), and CSF leak sometimes leading to meningitis. Much rarer complications include carotid artery injury and loss of vision. A recent variation of this approach has been to use an endoscope through the nostril to gain access to the pituitary transsphenoidally, thereby eliminating the need for conventional skin incisions and further improving postoperative recovery. The transsphenoidal approach is used in more than 90% of pituitary operations. It is appropriate for resection of microadenomas, enclosed macroadenomas with symmetrical suprasellar extension, and even some invasive adenomas. If the pituitary adenoma is very fibrous or exhibits significant extension into the middle cranial fossa, however, a transsphenoidal resection may be impossible, and it may be necessary to perform an intracranial operation.

The success of surgery alone in curing adenomas depends on the size of the tumor. In a review of transsphenoidal resection for GH-secreting adenomas, in most series hormonal normalization was seen in 67% to 91% of microadenomas but in only 48% to 65% of macroadenomas. ACTH-secreting adenomas, which generally are less than 1 cm at diagnosis, show a normalization of hormones following resection in 75% to 96% of cases in most surgical series. TSH-secreting adenomas are rare compared with other pituitary adenomas, and when diagnosed they are often large. Normalization of TSH after resection of these tumors has varied widely in different series, ranging from 33% to 86%.

The results of surgery for hormone-inactive adenomas are harder to document because the criteria for surgical success are less well defined. Clinical improvement can be seen even without total removal of the tumor. Series in which CT or MRI scans were performed a few months after surgery show a wide range of gross complete resection rates for these tumors, from 28% to 84%. A review of several surgical series showed that the likelihood of normalization of visual fields after surgery ranged from 16% to 53% and that visual field improvement occurred in 26% to 70% of cases.

Medical Therapy for Pituitary Adenomas

Although surgical resection and radiation therapy are effective therapies for prolactinomas, the primary treatment for these tumors currently is medical.  This is due to the availability of drugs that can suppress prolactin secretion and shrink these tumors. Surgical resection and radiotherapy are still used for the treatment of prolactinomas in the minority of patients who fail to respond to drugs or who cannot tolerate them.

Clinical manifestations of prolactinomas differ between the sexes. In premenopausal women, oligomenorrhea or amenorrhea and galactorrhea are extremely common. Infertility also may be the presenting sign, and women often have decreased libido. Estrogens have a marked stimulatory effect on prolactin synthesis and secretion; therefore, pregnancy can stimulate the growth of these tumors. In men and postmenopausal women, these tumors generally are asymptomatic until they are large enough to compress nearby structures, causing signs and symptoms such as visual deficits, headaches, and panhypopituitarism. Chronic hyperprolactinemia leads to decreased libido and impotence in 90% of men. Galactorrhea is uncommon in men but can occur in 10% to 20% of cases. The problems with reproductive and sexual function are due to inhibition of pulsatile gonadotropin secretion.

Secretion of prolactin by the lactotroph cells in the anterior pituitary gland is negatively regulated by dopamine produced by the hypothalamus. Therefore, dopamine agonists such as bromcriptine stimulate dopamine secretion by the hypothalamus and inhibit prolactin secretion by the lactotrophs. Administration of bromocriptine has been the standard therapy for prolactinomas for decades. This agent has been shown to decrease tumor size and normalize prolactin levels in 70% to 90% of patients. Bromocriptine also restores ovulation and menses and improves visual fields in a similar percentage of cases. Tumor shrinkage and decrease in prolactin levels can take anywhere from days to weeks to months to occur. Unfortunately, the action of bromocriptine is reversible; therefore, when the drug is discontinued, regrowth of the tumor with increase in the prolactin level usually is seen. Therefore, lifelong administration is the rule, and many patients can tolerate prolonged treatment for years. Bromocriptine has been used prophylactically during pregnancy to reduce symptomatic tumor enlargement. Of note, however, 10% to 20% of patients experience side effects such as nausea, vomiting, dizziness, postural hypotension, and headaches. In patients who have limiting toxicity with bromocriptine or whose tumor fails to respond to this drug, newer dopamine agonists such as lisuride, pergolide mesylate, cabergoline, and tergulide have been used with some success.

A few drugs are now available for the treatment of endocrine hypersecretion in pituitary tumors other than prolactinomas. For GH-secreting adenomas, somatostatin analogs such as octreotide and lanreotide have been shown to reduce GH levels in more than 90% of patients, with almost complete suppression in half. Synthetic agents that block GH binding to its receptor or growth hormone-releasing hormone to its receptors in the pituitary are being used experimentally. In general, these drugs are not being used for first-line therapy in patients with GH-secreting adenomas but rather are indicated in patients in whom surgery has been unsuccessful.

TSH-secreting adenomas often are very large at diagnosis, making complete surgical resection difficult. Because these tumors express somatostatin receptors, octreotide and lanreotide can decrease tumor size and decrease TSH secretion.

Radiation Therapy for Pituitary Adenomas

Radiation therapy also is highly effective in controlling pituitary adenomas. Nowadays, however, it is rarely given as sole treatment for newly diagnosed pituitary adenomas. In contrast with surgery, radiation therapy will not result in a rapid reversal of neurologic signs and symptoms or a rapid drop in hormonal secretion. Radiation therapy usually is reserved for patients who either have residual disease after surgery or have a recurrence after surgery. Radiation therapy occasionally is used as the sole primary therapy in patients who have a medical condition that makes their tumor inoperable.

The overall 10-year control rate using radiation therapy is on the order of 85% to 95% based on a number of large retrospective series.  In a study from the University of Heidelberg, in 138 patients with pituitary adenomas who received radiation as initial therapy or after recurrence, the overall local control was 95% with a mean follow-up time of 6 years. Likewise, in a study from the University of Florida with a median follow-up time of 9.2 years, the overall local control rate at 10 years was 93%. Ninety-eight patients in this series had surgery and radiotherapy as initial therapy, and their 10-year local control was 95%. This was comparable to the 10-year local control rate of 90% for the 23 patients who had radiotherapy alone for newly diagnosed adenomas, but better than the 80% control rate seen in 20 patients who received radiation for a recurrence after their initial surgery (P = 0.03). Similar findings regarding improved local control in patients receiving surgery and radiation “up front” compared with those receiving radiation at recurrence were obtained from the Princess Margaret Hospital study that examined 160 patients with hormonally inactive pituitary adenomas. Pituitary adenomas can occur in the pediatric population, although they are much less common than in adults. In one study in 11 patients aged 19 or younger with pituitary adenomas, treatment consisted of surgery plus radiation or radiation only. At a median follow-up time of 15.6 years, only two had failed to respond to treatment.

In the foregoing studies, local control refers to lack of disease progression, clinically and radiologically. In many patients who have hormone elevations at the outset, however, complete normalization may not be achieved after radiation therapy. In the University of Heidelberg series, of 68 patients with hormonally active pituitary adenomas, 52% of patients showed some reduction in their hormonal overproduction, but only 38% demonstrated complete normalization. Furthermore, in patients who had a response, it often took years, in some cases up to 9 years. A study from the Princess Margaret Hospital specifically analyzed data for 145 patients who received radiation for hormonally active pituitary adenomas. The progression-free survival rate was 96% at 10 years; however, the actuarial long-term biochemical remission rate was only 40%. Radiation is thus highly effective in preventing pituitary adenomas from growing; however, it is far less effective in normalizing hormone levels in patients with hormonally active tumors.

Late Effects after Pituitary Irradiation

In the foregoing series, the median doses ranged from 45 to 50 Gy. In the University of Heidelberg study, a statistically significant dose-response relationship was found in favor of a dose of 45 Gy or less. One of the worrisome potential late effects of using higher total doses is the possibility of radiation-induced optic neuropathy. In some series, visual problems develop in a few patients after radiation, presumably as a result of optic nerve damage, but such complications are uncommon, ranging in frequency from 0.7 to 2%.  As discussed earlier (under “Adverse Effects after Irradiation of the Brain or Spine”), the risk of optic nerve-chiasm injury is dependent on both total dose and dose per fraction. At the doses commonly used to treat pituitary adenomas (45 to 50 Gy), the risk of radiation-induced optic neuropathy with standard fractionation (1.8 to 2 Gy per day) is very low but not zero.

Doses of 45 to 50 Gy to the pituitary gland carry a substantial risk of causing hypopituitarism. In patients who did not have hormonal deficiencies at the start, the risk of developing insufficiency of a given hormone ranged from 10% to 30% in the series discussed earlier. It is likely that in half of all patients who received radiation doses of 45 to 50 Gy, deficiency of at least one pituitary hormone will develop 5 years after radiotherapy. This is an ongoing risk after radiation therapy, however, and can occur many years later; therefore, patients must be monitored indefinitely for this complication.

Second malignant neoplasms are always a concern in patients who receive radiation and who are expected to be long-term survivors. In an analysis of 334 patients with pituitary adenomas treated at the Royal Marsden Hospital with surgery and radiation therapy (median dose of 45 Gy), a secondary brain tumor (two astrocytomas, two meningiomas, one meningeal sarcoma) developed in 5 patients, for an actuarial risk of 1.3% at 10 years and 1.9% at 20 years. The relative risk for developing a second tumor compared with the incidence in the normal population was 9.3. A report from the Princess Margaret Hospital presented similar conclusions. In a series of 306 patients with pituitary adenomas treated with irradiation, gliomas of the brain developed in 4 patients, with a latency of 8 to 15 years. The relative risk compared with risk in the normal population was 16, with an actuarial risk of 1.7% at 10 years and 2.7% at 15 years.

Stereotactic Radiation Techniques for Pituitary Adenomas

A number of institutions have reported using stereotactic means of delivering radiation to pituitary adenomas, both in fractionated and single-dose regimens. Investigators have used FSRT with 45 to 50 Gy in 1.8-Gy daily fractions delivered using relocatable stereotactic headframes. Stereotactic radiosurgery also has been used, with single fractions ranging from 10 to 27 Gy. The local control rates using these techniques has been reported to be greater than 90%; however, the length of follow-up in these studies is too short to allow definite conclusions to be drawn.

The rationale for using single large doses is the contention that the interval to normalization of hormonal levels is shorter with this approach than with conventionally fractionated radiation therapy. Yoon and colleagues reported that 11 of 13 patients with prolactino mas had normalization of their hormone level within 1 year. Mitsumori and associates found that the average time to normalization with stereotactic radiosurgery was 8.5 months, versus 18 months with FSRT (45 Gy in 1.8-Gy daily fractions). Pouratian and associates reported on 23 patients with prolactinomas that had failed to respond to medical and surgical treatment and then were treated with Gamma Knife radiosurgery. Subsequently, a normal prolactin level was attained in 26%, by an average time of 24.5 months. Remission was significantly associated with the patient's being off a dopamine agonist at the time of irradiation. Box 4 summarizes the recommended approach to management of pituitary adenomas.

Box 4. 


  • Pituitary adenomas are extremely common as an incidental finding (in up to 10% of normal volunteers by magnetic resonance imaging screening).
  • These tumors may become symptomatic, when the patient comes to medical attention, because of hormone secretion, compression of nearby structures causing neurologic symptoms, or compression of the pituitary stalk, leading to hypopituitarism.
  • These tumors are classified by size as microadenomas (1 cm or less in diameter) or macroadenomas.
  • Initial therapy for most prolactinomas is with a dopamine agonist (e.g., bromocriptine, lisuride, pergolide), which usually decreases prolactin levels and shrinks the tumor.
  • Initial therapy for most other pituitary adenomas is transsphenoidal surgical resection, which is safe and leads to rapid reversal of neurologic signs and symptoms. Surgery normalizes hormone levels in most patients with microadenomas.
  • Radiation therapy currently is reserved for treatment of residual disease or recurrence after surgery, or for patients who are not eligible for surgery because of medical reasons. In patients with elevated hormones, normalization of levels after radiation therapy may take years. A dose on the order of 45 Gy in 180-cGy daily fractions should offer good control with extremely low risk of optic neuropathy.


Clinical and Pathologic Considerations

Acoustic neuroma has many other names; in addition to neuromas, these tumors also are referred to as neurilemmomas, neurinomas, neurofibromas, schwannomas, and nerve sheath tumors. They are benign tumors that most commonly originate from cranial nerve VIII, usually in the vestibular region of the internal auditory foramen, where the nerve acquires a Schwann sheath. For this reason, these tumors also are sometimes called vestibular schwannomas. Neuromas can affect other cranial nerves, such as the trigeminal nerve and nerves in the jugular foramen region; however, these are much less common than acoustic neuromas.

Acoustic neuromas account for 8% to 10% of all primary intracranial tumors, generally affecting people in the fifth decade of life. These tumors characteristically grow very slowly. Early on, they are asymptomatic, but with enlargement, they lead to progressive hearing loss and tinnitus. The hearing loss typically is in the conversational range. As the tumor expands and compresses cranial nerve VIII, it may cause vertigo and unsteadiness of gait. With growth into the cerebellopontine angle, cranial nerves V and VII may be compressed, resulting in otalgia, facial numbness, facial palsy, and change in taste. With continued growth, brainstem compression and obstruction of the fourth ventricle may develop, causing hydocephalus. Because signs and symptoms can arise insidiously, a progressive hearing loss and gait unsteadiness may evolve over years in some patients before the tumor is diagnosed.

Patients with NF-2 often have bilateral acoustic neuromas; in fact, this finding is diagnostic for NF-2. Acoustic neuromas in patients with NF-2 contain mutations in the NF2 gene and chromosome 22 deletions. Sporadic unilateral acoustic neuromas arising in patients without NF-2 also are associated with chromosome 22 deletion and NF2 mutation.

Surgery for Acoustic Neuromas

Traditionally, acoustic neuromas have been treated by surgical resection. A number of different approaches can be taken, including the suboccipital approach, the translabyrinthine approach, and the middle fossa approach. Some centers strongly favor one approach over the others, but many surgeons make a decision on a case-by-case basis. The translabyrinthine approach is the only procedure that inherently sacrifices hearing in the course of the procedure; therefore, it generally is not used in patients who have some residual useful hearing. Hearing preservation is possible but not guaranteed with either the suboccipital or middle fossa approach. In one series of patients who underwent a resection using the suboccipital approach with an attempt to preserve hearing, 18 of 46 (39%) patients who had good preoperative hearing maintained good hearing after surgery. The middle fossa approach is indicated in patients who have useful hearing and have tumors entirely contained within the internal auditory canal. In one series using this approach, total tumor removal was achieved in 98% of the cases, with hearing preservation in 59%. Approximately 89% of patients maintained normal or near-normal facial nerve function.

Catastrophic complications such as brainstem stroke, postoperative cerebellar hemorrhage, or death are rare after surgery for acoustic neuromas. Two other serious complications that are more common are CSF leak, which can lead to meningitis, and cranial nerve VII palsy. Some surgical teams feel that CSF leak and damage to the facial nerve are more likely with the suboccipital approach than the translabyrinthine approach; however, both complications have been described with both approaches. In an effort to decrease damage to the facial nerve, many surgeons routinely perform intraoperative electromyographic monitoring. Persistent headaches that last for months to a year are more common after surgery using the suboccipital approach, presumably as a result of aseptic meningitis from contamination of the subarachnoid space with bone dust when the internal acoustic canal is drilled intradurally.

Radiotherapy for Acoustic Neuromas

Radiation therapy often is effective in cases in which total resection cannot be performed. In a series from the University of California at San Francisco, postoperative radiation therapy (at a dose greater than 45 Gy) decreased the recurrence rate after subtotal resection from 46% (6 of 13 patients) to 6% (1 of 11 patients) (P = 0.01). Radiation therapy also is effective in controlling acoustic neuromas that are not surgically resected. Proton beam therapy has been used to deliver single large radiation fractions to acoustic neuromas. In the Massachusetts General Hospital series, 68 patients received a dose of 12 Gy to the tumor margin. With a median follow-up time of 44 months, the 5-year actuarial control rate was 84%.

Single large doses of radiation delivered using stereotactic radiosurgery also have been used. The University of Pittsburgh has accumulated one of the largest experiences with this approach. An analysis of their first 5 years of using Gamma Knife radiosurgery with a 12- to 20-Gy marginal dose showed a 98% local control. Of patients with normal function of cranial nerves VII and V before radiosurgery, however, 15% and 16%, respectively, developed some dysfunction afterward. Furthermore, of patients with useful hearing before radiosurgery, only 47% maintained the same level of hearing. Other institutions also found high local control rates, along with high rates of cranial nerve V and VII dysfunction, after single large stereotactic doses using either a Gamma Knife machine or a linear accelerator. On the basis of their initial results, the investigators at the University of Pittsburgh changed their policy by decreasing the marginal dose and using MRI scans for treatment planning purposes. From 1992 to 1997, 190 patients underwent Gamma Knife radiosurgery delivering 11 to 18 Gy to the margin of the tumor (median dose, 13 Gy). With a median follow-up time of 30 months, the 5-year actuarial tumor control rate was 97%, and 71% of patients retained useful hearing. The 5-year rates for developing facial weakness and facial numbness were 1% and 2.7%, respectively.

In order to potentially reduce late effects, a number of institutions have started using FSRT. Both hypofractionation (25 Gy in 5-Gy fractions, 30 Gy in 3-Gy fractions, 21 Gy in 7-Gy fractions) and conventional fractionation with 54 to 58 Gy in 1.8- to 2-Gy fractions or 36 to 44 Gy in 1.8-Gy fractions have been used. Although the follow-up is very short in some of these series, the control rates have ranged from 97% to 100%, with a useful hearing rate ranging from 72% to 85% and low rates of cranial nerve V and VII dysfunction. Although the jury is still out on use of single-dose stereotactic radiosurgery versus FSRT for acoustic neuromas, the experience at Jefferson University Hospital strongly favors the latter. Patients have received treatment with either Gamma Knife stereotactic radiosurgery (12-Gy marginal dose) or FSRT (50 Gy in 2-Gy fractions) in a nonrandomized fashion. The rates of cranial nerve V and VII preservation were equally high in both groups (93% to 98%), as were the local control rates (97% or better). A significant difference, however, was found in the rates of serviceable hearing (33% for Gamma Knife stereotactic radiosurgery versus 81% for FSRT).

As indicated by the foregoing findings, radiation therapy can be a useful alternative to surgery to control the growth of acoustic neuromas. Undoubtedly, however, debate will continue regarding the merits of one treatment over the other, as well as the optimal radiotherapy technique. Although the previous discussion centered on the treatment of acoustic neuromas, some evidence indicates that SRS is effective in controlling the growth of trigeminal neuromas and alleviating trigeminal neuralgia, as either an alternative or adjunct to surgery.


Clinical and Pathologic Considerations

Hemangioblastomas are low-grade vascular tumors that constitute 1% to 2% of intracranial tumors. They usually occur in the cerebellar hemispheres and vermis, although they can involve the pons, medulla, and spinal cord. Most cases occur sporadically, but 20% occur as part of the familial von Hippel-Lindau syndrome. Patients with this syndrome have a germline mutation in the VHL gene. Tumors that these patients develop have sustained a mutation in the second VHL allele, leading to loss of VHL protein function and resulting in stabilization of the HIF-1a protein under normoxic conditions. Stabilization of HIF-1a leads to constitutive high levels of expression of target genes including VEGF (vascular endothelial growth factor). For this reason, the tumors seen in patients with von Hippel-Lindau syndrome are highly vascular. Other abnormalities seen in these patients include retinal angiomas, renal cell carcinomas, and cysts involving many organs such as the pancreas, kidney, lungs, and liver. Pheochromocytomas also may develop in these patients, who often display erythrocytosis as a result of increased erythropoietin production. Sporadic hemangioblastomas, which occur in patients not suffering from von Hippel-Lindau syndrome, also contain mutations in the VHL gene in at least 20% of cases.

Patients with cerebellar hemangioblastomas often present in the third decade of life. Presenting manifestations stem from cerebellar dysfunction, increased ICP, and involvement of nearby cranial nerves. These signs and symptoms include headaches, nausea, vertigo, diplopia, tinnitus, ataxia, and poor coordination. The lesion is well visualized by CT scanning or MRI. Angiography, which generally is performed before surgery, shows the highly vascular nature of these tumors.

Histologically, hemangioblastomas show numerous capillary and sinusoidal channels lined with endothelial cells. A cystic component often is present; the cyst is filled with xanthochromic, proteinaceous fluid, with a vascular nodule in the cyst wall. In one series, 6 of 19 patients (32%) had a solid lesion without a cystic component. Lesions of this type are more likely to originate in the brainstem.

Therapy for Cerebellar Hemangioblastomas

The primary therapy for most patients with these tumors is surgical resection, if this can be done safely. If a cyst is present, it is drained; then the solid component is dissected and removed. If the tumor involves the brainstem, however, surgery may be extremely risky, with the potential for massive hemorrhage or extensive postoperative edema leading to death. Surgical resection historically has been associated with high local control rates. In one series, recurrence was noted in only 13 of 112 patients (12%) after surgery. In 6 patients, disease recurred in the original tumor bed, after incomplete removal, and in 10 patients, recurrence was in a different site or in the same site after gross total resection.