Chapter 40: Tumors of the Central Nervous System

Brain tumors are a heterogeneous group of lesions that range from benign, slow-growing tumors found only incidentally on autopsy, to malignant, rapidly growing tumors that cause death within months. The most common intracranial tumors are brain metastases from systemic cancer, estimated at 200,000 new cases per year in the United States, based on a 10% to 15% incidence (¹). In comparison, the incidence of primary brain and spinal cord tumors for 2014 was estimated at 23,380 new cases (American Cancer Society 2014 Facts and Figures [http://www.cancer.org/acs/groups/content/@research/documents/webcontent/acspc-042151.pdf (or, cancer.org]).

Because of the heterogeneous histology and often-refractory nature of primary brain tumors, their management is complex, ideally requiring a multidisciplinary team and individualized treatment. The diagnosis is made on the basis of histology, so an accurate characterization of the lesion pathology is crucial, often necessitating confirmation at a specialized cancer center. Optimal outcomes involve the coordination of neurosurgery, radiation oncology, and neuro-oncology specialists. Despite advances in neurosurgical techniques, radiation therapy, and chemotherapy, the prognosis for patients with high-grade gliomas such as glioblastoma (GBM), the most common form of glioma, remains dismal. Recent large clinical trials have reported a median survival of only 14 to 16 months with a 26% to 33% 2-year survival rate (^2,3). A review of eight consecutive phase II chemotherapy trials for recurrent GBM demonstrated only a 6% response rate (complete response [CR] and partial response [PR]), with a 6-month progression-free survival (PFS) of 15% and a 1-year survival of 21% (⁴). It is therefore important to consider patients with high-grade gliomas for entry into clinical trials at all stages of disease because new therapies target patients from initial diagnosis, with presurgical protocols, to salvage therapy at relapse. This chapter provides basic principles that can be used for diagnosing and treating patients with brain tumors along with an introduction to the molecular mechanisms underlying gliomagenesis.

Brain tumors are either primary tumors that arise de novo or secondary metastases, the latter being far more common. Most commonly, brain metastases result from lung cancer, followed by breast, melanoma, renal, and colorectal cancers. Most patients with brain metastases die from progression of their systemic cancer, although, because of improvements in systemic therapy, brain metastases are now seen more frequently and with increasing morbidity and mortality. On a more hopeful note, advances in treating brain metastasis with surgery and radiotherapy (RT) have improved overall survival when systemic disease is controlled.

Primary brain tumors are classified by the World Health Organization (WHO) grading system (Table 40-1), which is based on the histologic pattern of cell differentiation in the tumor, in addition to histologic features associated with biological aggressiveness (ie, mitotic figures, necrosis, vascular proliferation). Tumor grade is inversely correlated with prognosis. The most common primary brain tumors are gliomas (all glial tumors), followed by meningiomas, nerve sheath tumors, and pituitary tumors (⁵).

EPIDEMIOLOGY

Brain Metastases

Brain metastases occur far more commonly than primary brain tumors. Nearly any type of primary cancer can metastasize to the brain, including hematologic malignancies. Most commonly, brain metastases originate from lung cancer, which represents the second most common systemic cancer in men and women (⁶). The next most common pathology to metastasize to the brain is breast cancer, followed by melanoma, renal cancer, and colorectal cancer (⁷). Among solid tumors, melanoma has the highest propensity to metastasize to the brain. Based on autopsy findings, 40% to 60% of patients with melanoma develop brain metastases (⁸). Most brain metastases, particularly melanoma, present with multiple lesions, although when renal cancer metastasizes to the brain, it often results in a single lesion. Approximately 10% of metastatic lesions will present as an intraparenchymal hemorrhage. Hemorrhagic metastases are most commonly seen in patients with melanoma, renal cell carcinoma, thyroid carcinoma, and choriocarcinoma. The pattern of distribution of metastases in the brain varies depending on the primary cancer.

The incidence of brain metastases appears to be rising, which may be a consequence of increasingly sensitive imaging modalities such as magnetic resonance imaging (MRI), greater use of imaging, and the increasingly prolonged survival of patients with metastatic disease. The development of brain metastases usually occurs in the context of systemic relapse, although relapses can be isolated to the central nervous system (CNS). The impermeability of the blood–brain barrier (BBB), which often limits chemotherapy penetration into the CNS, may be the culprit in circumstances of isolated brain relapse.

Primary Brain Tumors

Gliomas are the most frequently occurring primary brain tumors and include astrocytomas, oligodendrogliomas (ODs), and ependymomas. Combined, these histologies account for approximately 40% of all primary brain tumors and over 80% of all malignant CNS tumors (⁹). The next most common tumor is meningioma (32%), followed by nerve sheath tumor (9%) and pituitary tumor (8%) (⁹). The most recent data from the Central Brain Tumor Registry of the United States reported an incidence of all primary benign and malignant CNS tumors of 21.03 cases per 100,000 for a total count of 326,711 incident tumors (¹⁰).

The incidence of primary brain tumors differs by age. Glioblastomas, which account for more than half of all gliomas, typically peak in incidence between ages 65 and 74, anaplastic gliomas peak between ages 45 and 54, and low-grade gliomas are most typically seen between ages 20 and 34 (⁵). In addition to the disparities of age, gender differences are seen in the incidence of primary brain tumors. The incidence of malignant brain tumors per 100,000 person-years in males is 7.7, compared with 5.4 for females (¹⁰) (Fig. 40-1).

Primary CNS lymphoma (PCNSL), concordant with AIDS, has decreased since its peak in the early 1990s, when it reached 10.2 per 1 million person-years. By 1998, the incidence decreased to 5.1 per 1 million person-years, which was attributed to the treatment of human immunodeficiency virus (HIV) with highly active antiretroviral therapy (HAART) in males under the age of 60 (¹¹). An increased incidence in PCNSL is seen not only in HIV/AIDS but also in iatrogenic immunosuppression, such as organ transplant, autoimmune disease, and cancer. The rate for persons over 60 years of age, however, has remained stable since 1994, at approximately 16 per 1 million person-years (¹¹).

Although many factors have been considered as putatively involved in gliomagenesis, therapeutic ionizing radiation is the strongest established causative agent underlying the development of brain tumors. Children with acute lymphoblastic leukemia (ALL), following prophylactic cranial irradiation, have an increased risk for both gliomas and meningiomas (¹²). Increased risk of brain tumors has been seen following therapeutic irradiation for pituitary tumors. Even low doses of radiation previously used to treat scalp ringworm increased the risk of developing nerve sheath tumors, meningiomas, and gliomas (¹³). Fortunately, diagnostic radiation does not appear to be strongly associated with the development of gliomas (¹⁴).

While a link between brain tumors and chemical exposure has been suggested, no specific agent has been identified with a link to brain tumors that can be validated with an exposure-disease correlation. No consistent link has been proven for the occurrence of cancer in agricultural workers. Positive correlations, but not causation, have been drawn between the occurrence of brain tumors and occupations involving exposure to synthetic rubber, vinyl chloride, and petroleum refining (¹⁵). Whereas smoking is implicated in an increased risk for many cancers, it is not associated with an increased risk of brain tumors. Exposure to cured foods has been linked to meningiomas, and nitrosamines have been associated with gliomas with a relative risk of 1.48 in adults with a high intake of cured meat (¹⁶). No correlation between alcohol intake or cosmetic use and brain tumors has been found (¹⁷).

There has been concern, particularly in the popular press, about the relationship between exposure to cell phones and the risk for brain tumors. Thus far, several case-control studies and a cohort study have failed to establish such a link.

Other than the increased incidence of PCNSL in patients with HIV infection, associations between viral exposures and brain tumors have not been consistent. Human cytomegalovirus (CMV) has recently garnered increasing attention in regard to its role in gliomagenesis. There is sufficient evidence that CMV sequences and viral gene expression exist in most gliomas and that CMV could modulate the malignant phenotype in GBM, but a specific role of CMV in glioma development has yet to be defined (¹⁸).

A protective effect of allergies (asthma, eczema, and hay fever) was noted in a meta-analysis of over 3,000 patients, with a relative risk of glioma incidence of 0.61 (¹⁹), supporting a role for immune modulation in brain tumor genesis. A retrospective case-control series demonstrated a decreased incidence of glioma in patients who reported histories of chickenpox, shingles, herpes simplex virus and Epstein-Barr virus (²⁰), and as a surrogate for exposure to infections in early life, birth order was correlated with increased risk for the development of glioma in adulthood (²¹).

Relatively few brain tumors are attributable to heredity; studies have cited from 1% to 5% (²²). Brain tumors can arise as a component of familial tumor syndromes, such as neurofibromatosis type 1 (NF1). Patients with NF1 are at increased risk for the development of gliomas originating predominantly in the optic pathway and brainstem. These neoplasms are typically pilocytic astrocytomas with a tendency toward a more indolent course in comparison to their sporadic counterparts. Neurofibromatosis type 1 is caused by a mutation of the NF-1 gene and is also associated with leukemia and pheochromocytoma. Neurofibromatosis type II is marked by a mutation of the NF-2 gene and is associated with bilateral vestibular schwannomas, meningiomas, and gliomas, including an increased risk for ependymomas. The Li-Fraumeni syndrome results from an autosomal dominant mutation of the p53 tumor suppressor gene, located on chromosome 17p. This p53 mutation results in many types of malignancies, including glioma and medulloblastoma, as well as sarcoma, breast cancer, leukemia, and adrenocortical cancer. Turcot syndrome is an autosomal dominant disease characterized by multiple polyps of the gastrointestinal tract as well as brain tumors. Two separate mutations have been identified in Turcot syndrome. One involves the APC (adenomatous polyposis coli) gene, which is associated with medulloblastoma. A mutation of the hMLH1 DNA mismatch repair gene is associated with GBM.

The understanding of cancer relies on uncovering the underlying molecular biologic mechanisms and signaling pathways that drive tumorigenesis. Delineating such mechanisms is complicated by the tremendous degree of intertumoral and intratumoral heterogeneity. Newer drugs target specific extracellular receptors or block intracellular signal transduction systems.

Glial Tumors

Pathway Alterations

Malignant glial tumors often exhibit significant histologic heterogeneity, which is reflected at the molecular level. Large-scale profiling efforts have accelerated our understanding of gliomas. The landscape of somatic genomic alterations has been comprehensively characterized by The Cancer Genome Atlas (TCGA). In the most recent publication by TCGA in 2013, alterations in three core overlapping pathways were described: receptor tyrosine kinase/RAS/phosphatidylinositol 3 kinase (RTK/RAS/PI3K) signaling (altered in 90% of GBMs), p53 signaling (altered in 86% of GBMs), and RB signaling (altered in 79% of GBMs). The alterations affecting the p53 pathway (MDM2, MDM4, and TP53); the Rb pathway (CDK4, CKD6, CCND2, CDKN2A/B, and RB1); and the PI3K pathway (PIK3CA, PIK3R1, PTEN, EGFR, PDGFRA, and NF1) were found to be mutually exclusive (²³). Of the 251 GBMs analyzed, at least one RTK was found altered: EGFR (57.4%), PDGFRA (13.1%), MET (1.6%), and FGFR2/3 (3.2%). Mutations of PI3Kinase were found in 25.1% of GBMs and were found to be mutually exclusive for PTEN mutations/deletions. The p53 pathway was found to be altered in 85.3% of tumors through mutations/deletion of TP53 (27.9%), amplification of MDM1/2/4 (15.1%), or deletion of CDKN2A (57.8%) (²³). The most recent TCGA analysis also identified mutations in genes for which targeted therapies have been developed, including BRAF (²⁴) and FGFR1/FGFR2/FGFR3 (²⁵).

The epidermal growth factor (EGF) pathway was found to be altered in 57.4% of GBMs in the most recent TCGA data. An alternate mutation in the external domain generates a truncated receptor, the EGFRvIII mutant, which is constitutively activated in gliomas (²⁶). Growth factors such as EGF and platelet-derived growth factor (PDGF) activate multiple signal transduction pathways that lead to cell survival and proliferation. The EGFR can activate the PI3Kinase pathway, which is also frequently mutated in glioma. When the PI3Kinase pathway is activated, the activation of AKT (protein kinase B) is triggered, in turn activating multiple prosurvival pathways, such as nuclear factor kappa B (NFκB), forkhead, and glycolysis. The activation of the PI3Kinase pathway has been associated with the reduced survival of patients with glioma (²⁷). Growth factors can also stimulate the ras pathway, which initiates a signal cascade through Raf/MEK/Erk and also promotes cell survival and tumorigenesis (²⁸). PTEN has phosphatase activity that inhibits the PI3Kinase pathway. The deletion of MMAC/PTEN leads to AKT pathway activation. The activation of p53 can result in either apoptosis or cell cycle arrest and initiation of DNA repair mechanisms. The abrogation of p53 activity would be expected to increase proliferation and mutations, leading to genetic instability, the prodrome of tumorigenesis.

Cell cycle regulation is a key target of carcinogenesis. Cell cycle checkpoints are affected by multiple proteins, acting either as accelerators or inhibitors of cell regulation. Important molecules in this process include p53, p21, and MDM. Another cell cycle regulation pathway important for glial tumors involves the retinoblastoma (RB) gene. When the RB gene is phosphorylated, the E2F transcription factor is released and activates cellular proliferation. The regulation of RB activity is complex and involves multiple cyclins (cyclin D), cyclin-dependent kinases (CDK4/6), and CDK inhibitors (p16), whose activities are under investigation (²⁹).

Gliomagenesis

The majority of GBMs (~90%) are primary, characterized by rapid development in the absence of any clinical or histologic evidence of a lower-grade lesion, in contrast to secondary GBMs, which progress from a low-grade astrocytoma or anaplastic astrocytoma (AA). Primary and secondary GBMs arise from distinctly different genetic pathways. Primary GBMs are postulated to develop quickly from glial progenitor cells over a period of a few months, potentially acquiring mutations in EGFR, TP53, or PTEN. In contrast, secondary GBMs are believed to arise from a stepwise accumulation of mutations.

Isocitrate dehydrogenase (IDH) mutations are believed to be a very early event in gliomagenesis that persist during progression to secondary GBM. The IDH1/2 mutations likely occur prior to the acquisition of a TP53 mutation, a driving force toward astrocytic differentiation, and prior to the acquisition of 1p/19q loss, believed to be the driving force toward oligodendroglial differentiation. Codeletion of 1p/19q has been classified as the genetic signature of ODs. Recent exomic sequencing has revealed that mutations in the CIC gene at 19q13.2 and FUBP1 gene at 1p are also frequently observed in ODs. In addition to the acquisition of TP53 mutations, ATRX mutations are often noted in WHO grade II and III astrocytomas and secondary GBMs (³⁰).

Predictive and Prognostic Factors

In anaplastic oligodendrogliomas (AODs), allelic losses of chromosomes 1p and 19q (1p19q LOH) have emerged as markers of chemotherapeutic response and longer survival (³¹), qualifying 1p19q loss as a prognostic factor. Recent studies have confirmed that 1p/19q LOH is also predictive of response to chemotherapy. In independent seminal studies by the European Organization for Research and Treatment of Cancer (EORTC) and Radiation Therapy Oncology Group (RTOG), patients with AODs harboring 1p/19q LOH did significantly better when treated with a combination of radiation and chemotherapy compared to treatment with radiation alone. Tumors without 1p/19q LOH did not incur a survival benefit (^32,33,34).

O6-Methylguanine-DNA methyltransferase (MGMT) is a DNA repair protein that reverses DNA damage induced by alkylating agents such as temozolomide and has been implicated as a major mechanism of resistance to alkylating agents. Hypermethylation of the promoter region of the MGMT gene, which inactivates MGMT gene transcription, has been associated with response to alkylating agents and increased survival in glioma patients (³⁵).

Isocitrate dehydrogenase mutations were first reported in 2008 (³⁶) and quickly became a seminal discovery. The IDH mutations now definitively distinguish primary from secondary GBM and have been established as a positive prognostic factor with an increase in overall survival noted in patients harboring an IDH mutation over those with wild-type IDH (³⁷). In low-grade glioma, anaplastic glioma, and GBM, IDH mutation status appears to be the most important prognostic factor (^36,38). In a study of 382 patients with high-grade gliomas, IDH1 mutation was found to be of greater prognostic relevance than histological diagnosis according to the current WHO classification system. The sequence of more favorable to poorer outcome was (1) AA with IDH1 mutation, (2) GBM with IDH1 mutation, (3) AA without IDH1 mutation, and (4) GBM without IDH1 mutation (³⁸).

Meningiomas

A mutation in the tumor suppressor NF2 gene on chromosome 22q12, has been closely associated with meningiomas and is disrupted in approximately half of meningiomas (³⁹). Germline mutations of this gene result in neurofibromatosis type 2, an autosomal dominant disorder that can manifest as multiple meningiomas, bilateral schwannomas, gliomas, and intracranial calcifications (⁴⁰). Merlin, the product of the NF2 gene, functions as a tumor suppressor gene and is a member of a family of cytoskeleton-associated proteins linked to RTK activity and ECM interactions (⁴¹).

In a recent study, whole-genome or whole-exome sequencing was performed on 17 sporadic meningiomas. The majority of meningiomas harbored simple genomes, with fewer mutations, rearrangements, and copy number alterations than observed in other adult tumors. Focal NF2 inactivation was confirmed in 43% of tumors. A subset of meningiomas lacking NF2 alterations exhibited recurrent oncogenic mutations in AKT1 and SMO with immunohistochemical (IHC) evidence of activation of their pathways. Mutations involving SMO were observed in 3/17 tumors, and AKT1 mutations were noted in 5/17 samples. Interestingly, these mutations were seen in the more therapeutically challenging tumors involving the skull base and were of higher grade (⁴²). Hyperactive Hedgehog (Hh) signaling has been linked to many other cancers and is under active investigation in meningiomas highlighting SMO as a potential therapeutic target (⁴³). Inhibitors of PI3K/AKT/mTOR are also under active investigation for meningiomas harboring AKT1 mutations.

Brain Metastases

The development of brain metastases is an intricate sequential process. In addition to proliferating, tumor cells must migrate and enter the systemic circulation, survive, travel/transport through the blood to the brain, adhere to and extravasate through the endothelium, invade the brain parenchyma, and proliferate, which requires the recruitment of a secondary blood supply. Failure at any of these steps will halt the metastatic process. Each of these steps requires complex interactions between the tumor cell and its changing microenvironment.

An understanding of the biological processes of brain metastases and the role of the BBB provides potential targets for intervention to improve treatment. There are multiple complex regulators of cell adhesion, including molecules such as integrins, cadherins, selectins, and heparin sulfate proteoglycans (⁴⁴). In addition, integrins can recruit intracellular signaling molecules such as focal adhesion kinase and src, which can lead to a cascade of cellular signaling that affects cell cycle control and proliferation. Integrins also play a part in regulating angiogenesis and tumor invasion. Other molecules that mediate invasion include the MMP family, serine proteases, and heparinase (⁴⁵).

Tumor cells must generate their own blood supply if they are to grow successfully and remain viable. Important activators of angiogenesis include vascular endothelial growth factor (VEGF), angiopoietin, hypoxia-inducible transcription factor (HIF), cyclooxygenase 2 (COX-2), PDGF, integrins, MMPs, and others. Important inhibitors of angiogenesis include angiostatin, endostatin, tissue inhibitors of matrix metalloproteinases (TIMPs), interferons, and platelet factor 4 (⁴⁶). Upregulation of activators or downregulation of inhibitors favors angiogenesis. As tumors grow, they begin to produce increasing numbers of angiogenic molecules that can participate in metastasis (⁴⁶). Because a multitude of pathways are involved in tumor growth and invasion, it is likely that if one pathway is inhibited, cells may escape through alternate pathways.

Effective drug delivery through the BBB to tumor cells is a significant obstacle to chemotherapy, both for brain metastases and primary brain tumors. The P-glycoprotein family of transporters actively exports drugs such as anthracyclines, Vinca alkaloids, taxanes, and etoposide (⁴⁷). The BBB can be breached by circulating cancer cells, which migrate across the BBB without degrading its permeability and proliferate. Once the tumor reaches a size that requires recruitment of new vessels, the BBB is disrupted, which allows imaging of brain tumors with contrast agents. In experimental models, brain metastases smaller than 0.25 mm in diameter are associated with an intact BBB, whereas larger tumors demonstrate BBB permeability (⁴⁸). Despite the presence of the BBB, studies of drug levels in brain tumors from systemic delivery have demonstrated pharmacologically relevant concentrations of drugs such as etoposide, cisplatin, cytarabine, and methotrexate. Measurements of drug levels in cerebrospinal fluid are not accurate indicators of tissue drug levels and also vary widely depending on the agent (⁴⁹).

Clinical Presentation

Brain tumors are usually diagnosed following presentation with symptoms such as seizure, headache, or focal neurologic deficits. We commonly see high-grade, malignant tumors presenting with headache, which reflects elevated intracranial pressure, and focal neurologic signs, such as weakness or aphasia. Low-grade glial tumors often come to attention with seizure, while other slow-growing tumors, such as meningioma, may be clinically silent and incidentally detected during imaging for an unrelated problem. We use contrast-enhanced MRI, the diagnostic standard, for brain tumor imaging. In addition to its superior sensitivity compared with computed tomography (CT), MRI provides more detailed anatomic as well as physiologic information that can contribute to a differential diagnosis. While contrast-enhanced CT can detect high-grade lesions that cause BBB breakdown, low-grade lesions may be detectable only on MRI, using sequences sensitive for edema and tissue changes. However, even in the case of known systemic primary cancer, contrast-enhanced CT may miss small foci of metastatic disease that are visible on MRI (Figs. 40-2,40-3,40-4,40-5,40-6,40-7,40-8,40-9).

The brain tumor imaging characteristics seen on MRI are helpful in making a diagnosis. However, confirmation of the diagnosis with pathology is necessary in nearly all cases. Noncancerous brain lesions that may be mistaken for malignancy include infection, demyelinating disease, vascular malformations, and stroke. A particular variant of demyelinating disease (tumefactive multiple sclerosis) with large focal tumor-like lesions, is known to resemble a malignant brain tumor. Unfortunately, some of these lesions have been irradiated, under the presumptive diagnosis of GBM, which only increases the severity of the demyelination. Conversely, patients with primary brain tumors are sometimes initially diagnosed with stroke or demyelinating disease. Further complicating the picture are patients who have brain tumors in addition to stroke, which is far more prevalent with increasing age. Patients with systemic cancer, often in remission, or in stable condition, can present with brain lesions that are suspected to be brain metastases but turn out to be a primary brain tumor. These types of cases often benefit from an interpretation by a specialized neuroradiologist who has been provided with a relevant patient history. A history of immunosuppression and multiple subcortical enhancing lesions may prompt the suspicion of a PCNSL or infection with toxoplasmosis. Further testing with brain thallium single-photon emission computed tomography (SPECT) scanning or fluorodeoxyglucose positron emission tomography (FDG-PET) imaging may serve to distinguish between the two possibilities.

Several classic radiographic appearances of brain tumors suggest malignancy. An irregular enhancing lesion with extensive edema following white matter pathways suggests a malignant glioma. Non–contrast-enhancing lesions with increased diffuse signals on FLAIR (fluid-attenuated inversion recovery) imaging suggest a low-grade astrocytoma. As a general rule for glial tumors, the presence of contrast enhancement suggests a high-grade malignancy. The WHO grade IV tumors nearly always enhance, as opposed to grade II tumors, which are typically nonenhancing. Two notable exceptions include pilocytic astrocytoma (WHO grade I) and pleomorphic xanthoastrocytoma (WHO grade II), which typically have an enhancing nodule and associated cyst. Meningiomas are typically homogeneously enhancing dural-based lesions associated with calcification. The appearance of multiple enhancing subcortical lesions with homogeneous enhancement suggests PCNSL. However, the same lesions, if associated with a known primary malignancy, may indicate brain metastases.

Diagnosis

The discovery of a brain lesion should prompt referral to a neurosurgeon for consideration of biopsy or resection. The management of brain tumors is critically dependent on a definitive pathology for diagnosis. A referral to a specialized neuropathologist may be necessary for diagnosis of uncommon or rare tumors and also in cases where there is only limited biopsy tissue available. Biopsies must be of adequate quality and be representative of the overall tumor to allow accurate diagnosis. Our institution insists on reviewing patient diagnostic slides prior to rendering a treatment recommendation, and it is not uncommon for our neuropathologists to disagree with the diagnosis provided by the referring physician. Primary brain glial tumors are graded according to the most malignant portion of the tumor. A brain lesion that is predominantly grade III astrocytoma but has a few regions that meet the criteria for grade IV astrocytoma (GBM) should be graded as a GBM. The tumor grade may be underestimated if the most malignant portion of the tumor is not sampled. Typically, the most malignant region corresponds to an area of contrast enhancement. If a patient is suspected of having PCNSL, the use of corticosteroids should be avoided. Primary CNS lymphoma can be sensitive to steroids which are lymphocytolytic during initial presentation, and the preoperative use of even small doses of corticosteroids can lead to a nondiagnostic biopsy. These patients usually require repeat biopsy after steroid discontinuation. Patients may present with deep central lesions involving the brainstem or thalamus. These cases may require referral to a specialized neurosurgical center to evaluate whether open biopsy, resection, or stereotactic-guided biopsy is appropriate. In these cases, close coordination with a department of neuropathology will be critical in obtaining adequate tissue for diagnosis.

Pathology

Diffuse astrocytomas are characterized by well-differentiated astrocytes—either fibrillary, gemistocytic, or rarely protoplasmic—with mildly increased cellularity. The cellular morphology of the tumor cells may differ within the same tumor sample and show great variability between tumors. Necrosis and microvascular proliferation are absent. Rare mitotic figures may occur, and nuclear atypia may be present, but not sufficiently to characterize the tumor as AA. The typical MIB-1 labeling index is less than 4% (Figs. 40-10 and 40-11).

Oligodendrogliomas are characterized by moderately cellular tumor cells with rounded, homogeneous nuclei, giving a “fried egg” artifactual appearance that is referred to as “classical OD.” A recent RTOG trial found that 80% of tumors with classical oligodendroglial morphology were associated with 1p/19q deletion, compared with only 13% of 1p and 19q deletions seen in nonclassical ODs (⁵⁰). Microcalcifications, microcyst formation, extracellular mucin deposition, and a dense network of branching capillaries are other oligodendrogliomal features. Nuclear atypia may be seen, but significant mitotic activity or microvascular proliferation is suggestive of an anaplastic tumor. The MIB-1 index is typically less than 5% (Figs. 40-12 and 40-13).

Anaplastic ODs are characterized by oligodendroglial cells, with signs of increased cellularity, nuclear atypia, and mitotic activity. Cellular pleomorphism may be present, with formation of multinucleated giant cells or spindle cells. Gliofibrillary oligodendrocytes and minigemistocytes are common. Although microvascular proliferation and necrosis may be present, their presence does not change the diagnosis to GBM. There is currently no designation for a WHO grade IV OD. The MIB-1 ratio is usually greater than 5% (Figs. 40-14,40-15,40-16.)

Anaplastic astrocytoma is characterized by diffusely infiltrating astrocytes with increased cellularity, nuclear atypia, and mitotic activity. They are more cellular than low-grade astrocytomas, and the nuclear atypia include formation of nuclear inclusions, multinucleated cells, and abnormal mitoses. Microvascular proliferation is absent; if present, it would upgrade the tumor to GBM. The typical MIB-1 labeling index ranges from 5% to 10% and can occasionally overlap with index values for low-grade astrocytoma and GBM (Figs. 40-17 and 40-18).

Glioblastoma and glioblastoma multiforme are synonymous; “multiforme” was dropped from the name in the 2007 WHO classification book, although glioblastoma is still commonly referred to as GBM. Glioblastoma is an anaplastic cellular tumor with marked nuclear atypia and mitotic activity, often with marked regional heterogeneity and cellular polymorphism. The presence of either microvascular proliferation or necrosis differentiate this lesion from AA. Other features associated with GBM include formation of epithelial “adenoid” structures, multinucleated giant cells, granular cells, lipidized cells, perivascular lymphocytes, and metaplasia. Glioblastoma is associated with a high proliferative rate, and MIB-1 labeling typically ranges from 15% to 20% (Figs. 40-19,40-20,40-21).

The accumulating evidence regarding the prognostic significance of IDH mutation status and gene expression profiling has highlighted the limitations of the current WHO criteria for the classification of diffuse gliomas. Molecular profiling is becoming an increasingly important tool in the separation of prognostic groups in diffuse gliomas. At our institution, molecular classification of diffuse gliomas has become standard. IDH and 1p/19q status are now routinely reported as part of an integrated diagnosis in addition to other pathologic features, including morphologic characteristics such as oligodendroglial or astrocytic phenotype, the results of other relevant IHC stains or diagnostic molecular tests (eg, p53, EGFR amplification, proliferation indices), and the method of molecular feature determination (eg, fluorescence in situ hybridization, IHC, comparative genomic hybridization) (⁵¹). This comprehensive format helps inform clinical decision making and patient counseling.

Meningiomas can have a wide range of appearances and are subtyped according to their appearance. Most meningiomas are grade I. The transitional variant has numerous concentric “onion bulb” structures. The psammomatous variant has calcified psammoma bodies. Pleomorphic nuclei and occasional mitoses are allowed, although four or more mitoses per 10 high-power fields would qualify in diagnosing atypical meningioma, grade II. Increased cellularity, high nuclear-to-cytoplasmic ratio, prominent nucleoli and foci of necrosis will also upgrade these to a grade II. Anaplastic meningiomas, grade III, have more than 20 mitoses per 10 high-power fields or obviously malignant cytology resembling carcinoma, melanoma, or high-grade sarcoma (⁵²) (Figs. 40-22 and 40-23).

Low-Grade Glioma

Diffuse astrocytomas are infiltrative low-grade brain tumors. Patients commonly present with new-onset seizures. Their peak incidence is in the third decade, followed by the second decade. Overall, these low-grade tumors represent 4% of glial tumors (⁹). The median survival time of patients with diffuse astrocytoma is between 5 and 8 years (⁵³). Variation in length of survival depends on patient age, performance status at diagnosis, and total versus partial tumor (as visualized on MRI) resection (^53,54).

The OD is a diffusely infiltrative, well-differentiated tumor composed of oligodendrocytes. These tumors comprise 3% to 4% of all primary brain tumors and approximately 7% of glial tumors, with an incidence of 0.3 per 100,000 per year. The peak incidence is from the ages of 30 to 50. Tumors with pure OD differentiation behave more indolently than astrocytomas. The ODs appear to be more sensitive to both chemotherapy and radiation therapy than astrocytomas, and the benefit from these therapies is more pronounced and durable. Median survival ranges from 4 to 12 years, and both OD and oligoastrocytoma (OA) have been included in the reported data (⁵³).

The OA is a mixed tumor composed of cells that resemble both OD and diffuse astrocytoma. Based on only a few cytogenetic studies, these tumors are thought to be of monoclonal origin. Clinically, these tumors present in a similar fashion to other low-grade glial tumors, which are best imaged with MRI and have imaging characteristics that indicate a diffuse, nonenhancing tumor. There has been no demonstration of a better prognosis with an increased proportion of the OD component. We treat these similarly to other low-grade glial tumors. Median survival ranges from 3 to 6 years. In general, these mixed tumors behave more aggressively than pure ODs but are possibly less malignant than pure astrocytomas (⁵⁵).

Clinical Management

Once a diagnosis of low-grade glioma is suspected, we recommend proceeding with a biopsy or resection to differentiate between a low-grade glioma and a nonenhancing anaplastic glioma (Tables 40-2 and 40-3).

If gross total resection or a major resection is possible without significant morbidity, neuro-oncologists generally recommend a complete resection, which may obviate the need for irradiation and decrease the risk of malignant transformation from residual tumor cells. In multiple retrospective series, total resection of nonenhancing tumor improves survival (⁵³). A volumetric analysis of the preoperative tumor, in addition to analysis of postoperative residual tumor, showed a correlation with time to recurrence and the likelihood of malignant transformation (⁵⁶). There are several low-grade tumors—such as pilocytic astrocytoma, ependymoma and subependymoma, pleomorphic xanthoastrocytoma, ganglioglioma, and dysembryoplastic neuroepithelial tumor—that may be definitively treated by complete surgical resection alone. These patients may benefit from treatment at a specialized neurosurgical center that treats a large volume of patients with brain tumors (Table 40-4 and Fig. 40-24).

If a complete resection is not recommended, therapeutic options range from observation to treatment with focal brain irradiation or chemotherapy. Older studies of low-grade astrocytomas suggested that radiation improved survival. The 5-year survival rates ranged from 49% to 68% for irradiated tumors compared with 32% for nonirradiated tumors (⁵⁷). The EORTC 22845 trial randomized patients to either up-front RT (54 Gy in 6 weeks) or delayed radiation therapy at progression. The PFS was 5.3 years in the early radiation group and 3.4 years in the control group. Overall survival, however, was 7.4 years in the up-front radiation group versus 7.2 years in the control group (²). No data were collected on quality of life. These data suggest that it is acceptable to delay radiation therapy until there are signs of tumor progression, especially for asymptomatic patients or patients who have had a complete tumor resection.

Patients with low-grade glioma should be carefully assessed to determine whether their symptoms are caused by the tumor. Formal neuropsychological testing may reveal cognitive deficits that are not apparent in the simple mental status screening used for dementia. In addition, these tests can be repeated to detect subtle cognitive decline, which may lead to a decision to alter a prescribed therapy. Patients who are symptomatic from their tumor from seizures, who have altered mental status due to tumor bulk or location, or who have other focal neurologic signs would be expected to improve with treatment of the tumor.

The etiology of cognitive decline in patients with a primary brain tumor is multifactorial. The causes include direct tumor effects from invasion and destruction as well as side effects from radiation therapy, chemotherapy, and anticonvulsants (⁵⁸). We have found the use of psychostimulants such as methylphenidate to be helpful in improving cognitive function, mood, and fatigue (⁵⁹).

Although there are concerns about the long-term effects of brain irradiation, radiation therapy is the current treatment standard (⁵⁸). The dose of RT currently used by the RTOG for low-grade glioma is 54 Gy to localized treatment fields as defined by the tumor appearance on T2-weighted MRI and including a 2-cm margin. A European trial involving 379 patients with low-grade glioma did not demonstrate a benefit for higher radiation dose when comparing 45 Gy with 59.4 Gy (⁶⁰). A second prospective study that randomized 203 patients with low-grade glioma to radiation therapy with either 50.4 or 64.8 Gy found a slightly lower survival (64% vs 72% at 5 years) and higher incidence of radiation necrosis in the group receiving the higher dose of 64.8 Gy (⁶¹).

Much less is known about the usefulness of chemotherapy for low-grade tumors. A small study of patients with incompletely resected tumors randomized to RT alone or RT with CCNU (lomustine) demonstrated a median survival time of 4.5 years with no difference between the two treatment arms (⁶²). An RTOG trial (RTOG 98-02) randomized patients with low-grade glioma and a high risk of recurrence (age ≥40 or subtotal resection/biopsy) to either radiation therapy alone or radiation therapy followed by six cycles of procarbazine, CCNU, and vincristine (PCV). The early results of this trial were presented at the American Society of Clinical Oncology (ASCO) meeting in 2014. The combination of PCV and radiation therapy prolonged both overall survival and PFS compared with radiation therapy alone in patients with high-risk grade 2 glioma, defined as patients with less than a subtotal resection and age over 40 years (ASCO abstract, J Clin Oncol. 2014;32:5s) (^62a).

The usefulness of chemotherapy as an initial treatment for patients with low-grade gliomas without high-risk features is unproven. The rationale for using chemotherapy as an alternative to radiation therapy for these patients is that although radiation therapy has a proven record of treatment response, it does not improve survival and may be associated with the significant long-term side effect of cognitive decline. It is hoped that chemotherapy will delay the need for radiation therapy without reducing treatment efficacy or survival (⁶³). Several limited studies have demonstrated an encouraging radiographic response by low-grade gliomas (primarily ODs but also astrocytomas) after treatment with temozolomide or PCV (^34,64,65). Patients who have residual tumor and would be at a high risk of cognitive side effects from radiation therapy may benefit from this strategy. However, there are no results from prospective randomized studies to recommend this approach. Patients with OD or OA may be more attractive candidates for applying this strategy, as they tend to have higher response rates to chemotherapy than patients with astrocytoma. The cytogenetic analysis of the tumor sample for LOH at 1p and 19q may predict patients who might benefit from this strategy.

Anaplastic Oligodendroglioma

The AODs comprise between 20% and 50% of all oligodendroglial tumors and approximately 5% of anaplastic tumors. The peak incidence is between ages 40 and 50. The clinical presentation of these tumors is similar to that of other anaplastic tumors, with focal neurologic signs, seizures, or symptoms of increased intracranial pressure. These lesions, which are usually contrast enhancing, can show calcification on CT scans as well as cystic structures, necrosis, and hemorrhage.

The initial standard therapy for an AOD is surgery, with the goal of gross total resection. In RTOG 9402, random assignment of 291 eligible patients with AO/AOA was made for the patients to receive PCV plus RT versus RT alone. There was no difference in median survival by treatment between the 148 patients randomized to PCV plus RT and the 143 patients randomized to RT (⁶⁶). However, the significance of codeletion of 1p/19q was supported by the results of EORTC 26951 with increased survival of both AOD and AOA with 1p/19q codeletion noted regardless of the treatment given: radiation alone versus radiation with chemotherapy (⁶⁷). The prognostic significance of 1p/19q was validated in the long-term results of RTOG 9402. Patients with co-deleted tumors lived longer than those with non–co-deleted tumors with the median survival of those patients with co-deleted tumors treated with PCV plus RT being twice that of patients who received RT alone. Neither timing (before, during, or following radiation treatment) nor dose intensity of PCV was found to be significant. No difference in median survival by treatment arm was appreciated in patients with non–co-deleted tumors (⁶⁶). A phase III study (CATNON) is under way to examine the appropriate treatment of anaplastic gliomas without 1p/19q codeletion.

Although PCV is the most studied regimen, in clinical practice temozolomide is typically favored for its more tolerable toxicity profile. Further study is ongoing from the RTOG/NCCTG/EORTC trial to determine if chemotherapy (PCV or temozolomide) can replace radiation and maintain the survival benefit. Despite initially high response rates, these tumors usually recur. Median survival for AOD treated with surgery, irradiation, and chemotherapy ranges from 3 to 5 years, although some patients survive past 10 years (⁵⁵). Recurrent disease is often treated with salvage regimens similar to those used for AA and GBM (Tables 40-5 and 40-6).

Anaplastic Astrocytoma

Anaplastic astrocytomas are diffusely infiltrating with nuclear atypia and anaplasia as well as marked proliferation, features that distinguish them from low-grade astrocytomas. A lack of vascular proliferation or necrosis distinguishes these tumors histologically from GBM. The highest incidence of AA is in the fourth decade, followed by the third decade, with nearly equal incidence rates in the second, fifth, and sixth decades. These tumors account for 7.5% of all glial tumors (⁹). Some patients may have a history of prior low-grade astrocytoma. Brain imaging shows diffuse hypointense tumor on CT scans and T1-weighted MRI. There is usually more mass effect and edema compared with low-grade astrocytomas, and contrast enhancement is typical. Because these tumors can occasionally be nonenhancing, neuroimaging alone is not sufficient to distinguish these lesions from low-grade astrocytomas. The median survival for patients with AA ranges from 5 to 7 years.

Optimal initial management begins with surgery with the goal of maximal, safe resection, both to provide adequate tissue for accurate analysis of pathology and to improve survival. Following surgery, limited-field radiation therapy to a target dose of 60 Gy is commonly recommended. The target radiation field typically includes the contrast-enhancing region of the tumor as well as the surrounding edema or nonenhancing tumor plus a 2-cm margin. The size of this field is often reduced after a 46-Gy dose has been applied to the contrast-enhancing lesion alone plus a 2-cm margin. Clinical trials using alternate radiation schemes of hyperfractionation or accelerated fractionation have not demonstrated an increased survival benefit over conventional fractionated conformal radiation therapy (⁶⁸). Adjuvant chemotherapy following radiation therapy increases time to progression and survival. Standard agents include combination therapy composed of PCV or (see Table 40-6) temozolomide (⁶⁹).

Patients with recurrent AA should be considered for clinical trials. Surgical resection should also be considered to provide a palliative benefit, relieve mass effect, allow dose reduction of steroids, and confirm histology. The recurrent tumor may actually have progressed to GBM from AA, and such patients are often eligible for a wider array of clinical trials than are available for recurrent AA. Trials have used temolozomide in combination with agents such as interferon alfa (IFN-α), cis-retinoic acid, metalloproteinase inhibitors, carmustine, irinotecan, and thalidomide (^70,71,72). Other agents that have been used for recurrent AA include tamoxifen, carboplatin, etoposide, irinotecan, and combination chemotherapy. To date, no single trial has proven to be superior (Fig. 40-25). Reirradiation can be considered for patients who are over 2 years beyond their original radiation treatment and for patients whose site of recurrent disease lies outside the initial radiation treatment field.

Glioblastoma

Glioblastoma is the most common and most malignant glial tumor of the brain. It comprises 50% of all glial tumors, with an incidence of approximately two to three per 100,000 per year (⁹). Glioblastomas are characterized by poorly differentiated astrocytes with cellular polymorphism, nuclear atypia, microvascular proliferation, and necrosis. The peak incidence is in the fifth decade, followed by the sixth and fourth decades. Glioblastoma is rare in children and young adults (⁹). Clinically, these tumors often present with signs of increased intracranial pressure, such as headache. They can also present with seizures or focal neurologic symptoms such as hemiparesis and aphasia, often with a short history of symptoms.

Imaging with CT or MRI usually reveals a contrast-enhancing lesion with irregular borders, frequently with a necrotic center. Vasogenic edema and nonenhancing tumor often surround the area of contrast enhancement and are best seen on T2-weighted or FLAIR imaging on MRI. Glioblastomas commonly spread through white matter tracts across the corpus callosum, internal capsule, and optic radiations. Multifocal lesions are seen. If these multiple lesions truly arise independently as opposed to spreading diffusely through tracts that are not visualized by imaging or pathology, they may have a polyclonal origin.

Glioblastomas are highly lethal. Despite extensive clinical research, survival has not changed greatly during the last 20 years. Prognostic factors include age and Karnofsky performance status (KPS). Surgical resection has shown some benefit, especially gross total resection, described when 90% or more of the enhancing tumor is removed (⁷³) (Fig. 40-26).

The addition of chemotherapy to radiation emerged as the standard of care for GBM based on the seminal large prospective, randomized, phase III trial from the EORTC. This trial randomized 573 patients to receive either standard RT (60 Gy in 30 daily fractions) or concurrent temozolomide (75 mg/m²/d) with RT followed by adjuvant temozolomide for 6 months (150 to 200 mg/m²/d for 5 days every 28 days). The group receiving concurrent and adjuvant temozolomide had a significant improvement in PFS (median 7.2 vs 5.0 months), survival (median 14.6 vs 12 months), and 2-year survival rate (median 26% vs 8%). Both groups had similar age, KPS, and surgical resection rates.

Our center strongly recommends patient participation in clinical trials, which enroll patients from initial resection to radiation therapy and salvage therapy at relapse. Patients are eligible for entry into a clinical protocol for recurrent disease if it has been greater than 12 weeks since completion of concurrent chemoradiation to avoid enrolling patients with pseudoprogression (radiographic change that can mimic tumor progression but is actually due to radiation-induced changes). If a patient is not enrolled in an “up-front” trial, we recommend evaluation by our neurosurgery service to explore the prospect of gross total resection. It is not unusual for our patients to have repeat resection of tumor following biopsy or subtotal resection at an outside institution. Following resection, we treat patients with concurrent temozolomide (75 mg/m²/d throughout radiation therapy) and standard conformal radiation therapy (59.4 Gy in 1.8-Gy fractions). Following radiation therapy, we use adjuvant temozolomide or temozolomide combination therapy. Although the EORTC study only used adjuvant temozolomide for 6 months, we typically continue treatment for at least 1 year, given the lethal natural history of GBM.

Dose-dense scheduling of temozolomide was evaluated in a large randomized, phase III trial based on the premise that prolonged exposure to temozolomide would result in prolonged depletion of MGMT, possibly translating into an improved survival in patients with newly diagnosed GBM. Standard adjuvant temozolomide (days 1-5 every 28 days) was compared to a dose-dense schedule (days 1-21 every 28 days). No statistically significant difference in either median OS or median PFS was observed between the two treatment arms. Treatment toxicity was higher with the dose-dense schedule (³).

Patients with GBM and progressive disease are offered salvage therapy if their KPS is adequate. We consider options including resection of tumor, chemotherapy, and stereotactic radiation therapy. Some novel neurosurgical clinical trials have offered local therapy with gene therapy using p53, although this was limited by lack of dispersion of the therapy into surrounding tissues (⁷⁴). An interleukin 13–conjugated Pseudomonas exotoxin has been studied using convection-enhanced delivery to lead to higher tissue concentration with larger volumes of distribution in phase I (^75,76). Another ongoing trial uses a conditionally replication-competent adenovirus (Delta-24-RGD) injected into the resection cavity for recurrent malignant gliomas.

One advantage of re-resection of progressive disease is to confirm pathology and specifically to determine whether the progressive enhancement on MRI represents tumor or radiation necrosis. Magnetic resonance imaging dynamic contrast and MR spectroscopy imaging, FDG-PET scanning, and brain SPECT thallium imaging sometimes help to distinguish between these two possibilities. However, all of these modalities have limited sensitivity and specificity, and sometimes the pathology reveals both treatment-related necrosis and foci of active tumor. Patients with pathology-confirmed radiation necrosis are often treated with steroids. More recently, bevacizumab, a monoclonal antibody targeted against the VEGF, has been utilized to treat radiation necrosis (⁷⁷).

Chemotherapy for recurrent disease typically produces response rates less than 10% and a 6-month PFS of 15% (⁴). Response rates that include stable disease and complete or PRs are 40% at best, but as the 6-month PFS value indicates, these responses are not durable. It is hypothesized that the multiple mutations and alterations in GBM and the heterogeneity of the tumor cell population may partially explain the striking resistance of these tumors to therapy. Younger patients respond best to chemotherapy, although responses to alkylating agents can be seen in patients older than 60 years of age. Long-term survivors of GBM (over 5 years) have typically had gross total resection, radiation therapy to a dose of 60 Gy, and chemotherapy, generally with temozolomide or a nitrosourea or other alkylating agent.

Salvage agents used for malignant glioma are identical to those used for recurrent AA (see Table 40-6). Rechallenging with continuous dose-intense temozlomide 50 mg/m²/d is a valuable therapeutic option as evidenced by the RESCUE study. The overall 6-month PFS for recurrent progressive GBM was 23.9% (⁷⁸) in contrast to 15% based on a pooled analysis of eight consecutive phase II trials of cytostatic and cytotoxic agents. In this study, the greatest therapeutic benefit was observed in patients with progressive disease during the first six cycles of conventional adjuvant temozolomide therapy (150-200 mg/m² × 5 days every 28 days) or after a treatment-free interval (⁴).

Bevacizumab has been approved by the Food and Drug Administration (FDA) for progressive disease following prior therapy, based on two trials. One study showed a 6-month PFS of 42% and overall survival of 8.7 months in patients receiving bevacizumab alone (⁷⁹). Another study showed a response rate of 19.6% with median duration of 3.9 months. The 6-month PFS was 29%, and 6-month survival was 57%. In addition, 50% of patients experienced decreased cerebral edema, 58% were able to decrease corticosteroid dependency, and 52% had improvement in neurologic symptoms (⁸⁰).

Bevacizumab was subsequently evaluated in phase III clinical trials for newly diagnosed GBM. Unfortunately, no effect was seen on overall patient survival. Two recent large randomized, phase III trials, AVAglio and RTOG 0825, demonstrated that the addition of bevacizumab to up-front treatment with radiation and temozolomide conferred no benefit in terms of overall survival. Progression-free survival was prolonged in both studies by approximately 3 to 4 months, reaching statistical significance in the AVAglio study but not in the RTOG 0825 study based on predefined criteria (^81,82).

Other active agents include irinotecan and carboplatin, which have been investigated as single agents in the salvage setting and in combination with bevacizumab in bevacizumab-naïve recurrent GBM (⁸³). The optimal schedule and combination of bevacizumab with alternative drugs has not been identified. Agents targeting angiogenesis have also been studied, although the use of interferon, thalidomide, EGF-RTK antagonists, and integrin receptor antagonists is not standard. Many of these targeted therapies demonstrated only limited activity as single agents, and efforts are under way to combine them with cytotoxic therapy (⁸⁴). Other cellular pathways being investigated with small-molecule inhibitors include the ras pathway with farnesyl-transferase inhibitors and the PI3Kinase pathway with mTOR inhibitors. Other novel approaches to malignant brain tumor therapy include use of oncolytic adenovirus, vaccine and dendritic cell immunotherapy, and histone deacetylase inhibitors. An important reaction has been discovered from the interaction between anticonvulsants that induce the hepatic cytochrome P-450 3A4 enzyme and other chemotherapy agents also metabolized by this same enzyme. Pharmacokinetic studies of patients with malignant glioma on single-agent irinotecan and sirolimus found significantly lower levels of active drug in patients on enzyme-inducing anticonvulsant drugs (⁸⁵). We recommend that patients on chemotherapy avoid the use of anticonvulsants that induce the expression of the P-450 3A4 enzyme whenever possible (Table 40-7).

Seizure Control

The management of seizures in patients with brain tumor is important to improve patient functioning and quality of life. A decline in seizure control may indicate tumor progression or worsening edema. It may also indicate a systemic infection or a drug interaction leading to decreased anticonvulsant drug effectiveness. In the case of tumor progression, a reduction in the amount of brain edema with high-potency corticosteroids (dexamethasone) may be sufficient to prevent further seizures. A second anticonvulsant is often necessary. Dexamethasone is a hepatic cytochrome P-450 3A4–inducing agent and often causes a reduction in serum levels of antiepileptic medications such as phenytoin and carbamazepine (also enzyme inducers) when the dose is increased. Similarly, patients can become symptomatic with toxic levels of anticonvulsants in the midst of a dexamethasone taper. It is important to follow serum anticonvulsant levels when using agents metabolized by the cytochrome P-450 system. Anticonvulsants that are highly protein bound can demonstrate significant changes in levels of circulating free drug without significantly changing the total serum level. It is useful to check serum-free phenytoin or valproic acid levels when patients taking these agents have seizures or show signs of toxicity.

Despite the numerous choices of anticonvulsants, it can be difficult to control seizures. Of the newer generation of anticonvulsants, we have had success using levetiracetam and lacosamide, which are easily titrated without significant drug interactions. Phenobarbital and clonazepam can be useful in resistant cases of seizures. Short-term use of lorazepam can help bridge changes in anticonvulsant regimens.

Quality-of-Life Considerations

It is critical to provide effective supportive care to patients with brain tumors to improve their functional status and quality of life for themselves and their caregivers. This care is typically labor intensive and often beyond the means of patients and their families to provide. We involve social work and case management early in the treatment of patients. They can provide interventions that may prevent a later breakdown in care.

The incidence of depression is high among this population and should be treated early. The causes of depression are typically multifactorial and may include direct effects of the tumor, side effects of chemotherapy and radiation therapy, and side effects of steroids in addition to issues associated with a loss of independence and a diagnosis of cancer. We suggest referral to psychiatry to optimally address these issues. A related concern is the impact of fatigue and somnolence, common side effects of brain radiation. We advocate the use of psychostimulants such as methylphenidate to treat both fatigue and cognitive side effects (⁵⁹). Although there are theoretical concerns that the use of stimulants may exacerbate seizures, we have not observed this in practice.

Patients often require high doses of steroids to manage edema and experience both acute and chronic toxicities from their administration. Acutely, the steroids may induce hyperglycemia requiring an insulin sliding scale. Patients often become agitated and irritable, suffer extreme mood swings, and even become psychotic when taking steroids. Low-dose neuroleptics can be effective in treating these side effects. Clinicians should aim to taper steroid use to the lowest doses necessary. Patients typically tolerate initial steroid weaning but often experience fatigue or worsening of neurologic function as dexamethasone doses are reduced to below 4 mg daily. This can be ameliorated with an extremely slow steroid taper and only lowering doses every 1 to 2 weeks by decrements of 1 mg or even 0.5 mg. Psychostimulants can help treat the inevitable fatigue experienced with the steroid taper. There is no effective treatment for steroid myopathy other than tapering off steroids and initiating physical therapy and rehabilitation as early as possible.

Meningioma

Meningiomas comprise 32% of primary brain tumors, with a rate of 5.35 per 100,000 person-years. The incidence of meningioma increases with age; the median age at diagnosis is 64 years (⁹). The tumor is often discovered incidentally without any symptoms. Clinically, these tumors typically present with headache, cognitive or personality changes, persistent focal neurologic deficits, and sometimes seizure. The main treatment is surgical resection, with the goal of complete resection when possible, accounting for relative risks and benefits depending on the patient’s age and condition (⁸⁶). Options for residual tumor include observation and radiation therapy, which can incorporate stereotactic delivery to minimize effects to local tissue (⁸⁷).

Chemotherapy for meningioma has been used for patients who have progressive disease after resection and RT; it is sometimes used adjuvantly following RT when pathology indicates malignant meningioma. Response rates have been disappointing in small case series. Agents that have been used include hydroxyurea (⁸⁸), IFN-α (⁸⁹), and liposomal doxorubicin (⁹⁰). Results using temozolomide have been discouraging, with no responders (⁹¹). As there are no established treatments after surgery and radiation have been exhausted and response to chemotherapy has been disappointing, the use of molecularly targeted therapy is being explored in aggressive meningioma. Frequently, EGF, PDGF, and VEGF receptors are overexpressed in meningiomas. Clinical trials using small-molecule signal transduction inhibitors such as erlotinib, gefitinib, and imatinib are being explored but have not yet shown significant efficacy (^92,93). In a recent phase II trial, sunitinib, a small-molecule tyrosine kinase inhibitor that targets VEGF and PDGF receptors, was found to have activity in patients with recurrent atypical/malignant meningiomas and warrants investigation in a randomized trial (⁹⁴). A novel genomic-driven clinical trial currently in development will examine the efficacy of SMO and AKT inhibitors in patients with surgery-confirmed mutations in these oncogenes.

Primary Central Nervous System Lymphoma

In contrast to most other brain tumors, chemotherapy is the initial treatment of choice for CNS lymphoma. Efforts at surgical resection have largely been discouraged as PCNSL has a tendency to involve deep brain structures and a multifocal pattern of growth. The traditional view has been that gross total resection conferred no survival benefit over biopsy, but this view has been challenged recently. In a phase III trial of 526 patients, PFS was found to be significantly shorter in patients who were biopsied compared to patients who had undergone subtotal or gross total resections, with no difference in outcome attributable to KPS or age, suggesting that surgical resection could be considered for patients with single lesions for which resection is deemed safe (⁹⁵).

Methotrexate-based, multiagent chemotherapy has been viewed as the treatment of choice in PCNSL. The incorporation of high-dose methotrexate (greater than 1 g/m²) has resulted in a significantly greater response and improved survival compared with previous regimens using a CHOP regimen prior to whole-brain radiation therapy (WBRT). A report by DeAngelis, incorporating methotrexate (1 g/m²), followed by whole-brain irradiation and two cycles of high-dose cytarabine (ara-C) (3 g/m²), demonstrated a median survival of 42.5 months (⁹⁶). This strategy is the basis for current CNS lymphoma protocols that have increased the dose of methotrexate and incorporated agents that more easily cross the BBB, such as procarbazine. A follow-up clinical trial incorporating methotrexate at 3.5 g/m² with procarbazine and vincristine, followed by whole-brain irradiation and cytarabine demonstrated a median survival of 60 months (⁹⁷). The improvement in patient survival has also brought to attention significant rates of cognitive decline and radiation-induced dementia, especially in patients older than 60 years (⁹⁸).

Current approaches to therapy of CNS lymphoma are investigating whether radiation therapy can be avoided or delayed to reduce cognitive decline and dementia without adversely affecting survival. Preliminary results from a trial using single-agent methotrexate at 8 g/m² every 2 weeks demonstrated a PFS of 12.8 months. Median survival had not been reached at more than 22.8 months (⁹⁹). Many clinicians at our center are cautiously delaying radiation therapy until relapse and continuing to use high-dose–based methotrexate regimens. In hoping to improve the results of single-agent methotrexate (⁹⁹), some regimens continue to incorporate procarbazine. Other agents that may be active in this setting include temozolomide and rituximab.

Patients with recurrent disease may respond again to methotrexate. Other regimens used include PCV (¹⁰⁰), high-dose cytarabine (¹⁰¹), temozolomide (¹⁰²), rituximab (¹⁰³), and the combination of temozolomide and rituximab (¹⁰⁴). High-dose chemotherapy with autologous stem cell rescue may also be effective (¹⁰⁵).

Brain Metastasis

The treatment of brain metastasis involves optimal interactions between oncology, neurosurgery, and radiation therapy. Depending on the setting of relapse, patient survival may depend more on local tumor control in the brain or on systemic control for progressive metastasis. Advances in local brain tumor control with surgery and radiosurgery will not improve patient survival if the patient ultimately succumbs to progressive systemic disease or continues to develop new brain metastases. The median survival of patients with brain metastases is 3 to 6 months (¹⁰⁶).

The options for therapy include surgical resection, WBRT, stereotactic radiosurgery (SRS), and systemic chemotherapy. Surgical resection is the treatment option considered primarily in patients with single large tumors. Resection of brain metastases has emerged as a standard treatment option for patients with surgically accessible single lesions, good performance status, and controlled or absent extracranial disease. The advantages of surgical resection is that the mass effect can be immediately ameliorated and removal of the tumor decreases edema. Surgical resection also provides pathologic confirmation of the diagnosis. Stereotactic radiosurgery utilizes multiple convergent beams to deliver a single high dose of radiation to a discrete target volume and is usually reserved for lesions whose maximum diameter is 3 cm or less. The ability to treat locations that were otherwise considered surgically inaccessible is a distinct advantage of SRS. Whole-brain radiation therapy via 30 to 40 Gy (in daily fractions of 2 to 3 Gy) is the standard therapy for brain metastasis, with an established body of literature supporting its use for multiple metastases. This therapy has the ability to eradicate micrometastatic disease to delay recurrence (^106,107) and is often used in conjunction with surgical resection or radiosurgery. It is well tolerated and can be effective for radiosensitive tumors such as metastases from small cell lung cancer or germ cell tumors. The greatest concern about WBRT has been the risk of neurocognitive effects, which can range from mild impairment to dementia. Hippocampal-sparing WBRT has been under active investigation in a large multicenter clinical trial (clinicaltrials.gov, identifier NCT01227954) as a technique to reduce neurotoxicity. Overall, the neurocognitive impact of WBRT in patients with brain metastases has not been well studied, and future research efforts will focus on the identification of risk factors predicting vulnerability. Neurocognitive end points should also be integrated into clinical trial designs.

We frequently treat patients at our institution with surgery if the lesions are greater than 3 cm and the patients are symptomatic. If the patient’s medical condition makes a surgical procedure risky, the patient may receive WBRT. Patients who have lesions smaller than 3 cm can receive radiosurgery if they are asymptomatic or if the lesion is in a deep region not amenable to resection. However, patients with symptoms resulting from the lesions more frequently receive surgery to remove mass effect as long as their medical condition permits. There is also debate over the role of whole-brain irradiation following surgery or radiosurgery to single lesions.

Chemotherapy for Brain Metastasis

Several small clinical trials and case reports support the concept that systemic chemotherapy demonstrates activity in treating brain metastases. Chemosensitive tumor types include breast cancer, small cell lung cancer, and germ cell tumors. The primary consideration in choosing a given regimen of chemotherapy is to use agents with known activity in a given tumor type. In many trials, response rates of brain metastases have been comparable to response rates of systemic disease. Patients who have had prior chemotherapy usually respond at lower rates.

Most clinical trials of investigational agents for solid tumors explicitly exclude patients with brain metastases. Compounding this omission is their common inclusion in studies of a heterogeneous group of patients with mixed tumor types and differing prior exposures to chemotherapy. Patients might also be expected to be more resistant to treatment with chemotherapeutic agents if they had failed RT. If chemotherapy is given during and after RT, it may be difficult to separate the efficacy due to RT versus chemotherapy. These factors make it difficult to compare treatment regimens and interpret studies (¹⁰⁸).

Newer drugs targeting specific extracellular receptors or blocking intracellular signal transduction systems are under investigation. Owing to their specificity, they often lack the side effects commonly associated with standard cytotoxic chemotherapy. If the therapeutic target is crucial for the cancer cell’s continued viability, the drug can be especially effective. The identification of BRAF mutations in 50% to 60% of advanced melanomas resulted in the development of potent and selective inhibitors. Vemurafenib and dabrafenib are FDA approved for the treatment of advanced melanoma and have transformed melanoma therapy, with high response rates seen in patients even with advanced, symptomatic, metastatic disease (¹⁰⁹).

Immune checkpoint blockade is emerging as a highly effective immunotherapeutic strategy in metastatic melanoma and other solid tumors. Ipilimumab, a human immunoglobulin (Ig) G1 monoclonal antibody to cytotoxic T-lymphocyte antigen 4 (CTLA-4), has been demonstrated to result in a durable response and improved overall survival when compared to non–ipilimumab-containing treatment arms in randomized trials (¹¹⁰). Ipilimumab has also been shown to be efficacious in the treatment of patients with melanoma with brain metastases, as evidenced by the measurable tumor reduction seen with ipilimumab used as monotherapy (¹¹¹).

The use of chemotherapy for brain metastases is faced with great challenges. The most important imperative is to discover new agents that can overcome tumor resistance to standard chemotherapy, whether through selection by prior pretreatment or inherent chemoresistance of tumor cell clones that metastasize from a primary site. Because most patients with brain metastases succumb to progressive systemic disease, improvement of local brain control will likely have a limited effect on survival. Conversely, the development of agents that are effective in establishing durable tumor control, both systemically and in the brain, will improve survival, as in the unique case of germ cell tumors.

Clinical variables associated with survival in the setting of brain metastases have been studied. The most well-known prognostic scoring system is the recursive partitioning analysis classification developed from 1,200 patients who received WBRT in the RTOG database. Patients were categorized into one of three classes based on age, KPS, status of primary tumor, and extent of extracranial disease (¹⁰⁶). Recently, the Graded Prognostic Assessment scale was developed based on the analysis of 1,960 patients in the RTOG database; it also incorporates the number of metastatic lesions in the scoring system (¹¹²) These prognostic scoring systems may help identify patients who might benefit from chemotherapy and help design clinical trials that account for specific tumor histology and prior exposure to chemotherapy. Improvement in patient survival will result from improved local control of CNS disease if the primary disease site remains dormant, illustrating the need for a multimodality approach to the treatment of the patient with brain metastases.