Clonal Expansion of Malignant Tumors

A fundamental concept in cancer is the clonal expansion of tumors, with each step in tumor formation based on adding another mutation to the tumor genome. Figure 102-1A illustrates this concept. The process starts with a single mutation in a single cell. As mutations in critical genes accumulate in a cell, a cancer develops in stepwise fashion. In brain cancers, it is not clear how long it takes for mutations to accumulate and in what order all the mutations might occur because the early stages of the tumors are not easily observed during tumorigenesis. It appears, however, that many mutations and changes in copy number, in any one of many different combinations, accumulate to form a GBM.⁷ Because this clonal expansion presumably arises from a single cell, all the cells in the tumor should have the same mutation. This is normally true, except in a few situations, such as the presence of unstable amplifications that give rise to heterogeneity for this change in the tumor or the acquisition of a late-occurring mutation in the latter stages of tumor cell clonal expansion. For the vast majority of mutations, however, the same mutations will appear in all parts of the tumor.

FIGURE 102-1 Basic concepts in cancer genetics. A, Clonal expansion of tumors. Cancers are thought to start with a single mutation in a single cell and then accumulate the necessary mutations in critical genes to grow in an uncontrolled manner. Each of the cells in the tumor will therefore have the same complement of mutations and will be “clones” with respect to mutation sequences. In this diagram, progressive mutations are shown by darker shades of pink. Tumor growth may not occur appreciably in many tumors until there are numerous mutations. In this diagram only three steps are shown, although for the highest grade malignancy many more mutations per cell are observed. B, Common mechanisms of gene mutations and alteration. Oncogenes are activated with a gain of function, whereas tumor suppressors are inactivated. Point mutations and other small intragenic sequence changes can activate or inactivate a gene. Entire genes can be lost or gained with homozygous deletion and genomic amplification. Parts of two genes can be fused during chromosomal amplification to form a new oncogene.

Inherited Mutations and Familial Syndromes

Cancer genetics started with identification of the inherited gene mutations responsible for familial clustering of cancer through linkage or association studies. Linkage and association studies can locate the approximate region of the chromosome responsible for a disease phenotype based on finding a polymorphic marker that cosegregates in families with individuals affected by the disease. These studies of inheritance are laborious because they involve identifying large families with the disease, collecting blood from family members, and studying markers throughout the genome to find one that is on the same chromosome, close enough to the gene of interest so that the marker and mutant gene are not separated by DNA crossover during meiosis.

The prototype cancer gene syndrome involves the inheritance of a tumor suppressor mutation in which there is one mutated allele in the germline and a second spontaneous mutation inactivates the second allele. Other spontaneous mutations occur subsequently in the cell as part of the cell’s transformation to a malignant state. This first hit dramatically increases the chance of tumor formation in someone with the germline mutation, but by itself it is not sufficient.

Because the first hit is inherited as a germline mutation, it can be transmitted to offspring. Although most malignant tumors observed in the clinic do not have an evident hereditary basis, it is important to recognize the possibility of familial clustering of brain tumors and consider genetic counseling and further evaluation if evident. Table 102-1 lists some of the more common cancer-associated mendelian disorders that have brain tumors as part of the phenotype. Mendelian disorders are genetic diseases that have a clear pattern of inheritance within families, such as dominant or recessive inheritance. Most of the syndromes that involve brain tumors have an autosomal dominant mendelian inheritance pattern. A complete catalogue and literature review of all the mendelian disorders identified is readily accessible at the Online Mendelian Inheritance in Man (OMIM) website.⁸

In the laboratory, the key way to distinguish a somatic mutation from a hereditary mutation is to sequence the suspected gene mutation in both the tumor and normal tissue. In most cases, lymphocyte DNA from a blood sample is used as the normal control. If the mutation appears only in the tumor and not in normal tissue, it is evidence for a somatic mutation acquired during tumor development.

The classic scenario for a brain tumor phenotype with mendelian inheritance is the dominant inheritance of a mutation in a tumor suppressor gene, with an average of 50% of offspring inheriting one copy of the mutated gene from the affected parent. In the case of tumor suppressor genes, both copies need to be inactivated to initiate the process of tumor formation. When one mutation is inherited as the first “hit,” there is a high probability of the second being inactivated. Inheritance of this first hit greatly increases the risk for cancer.

To find evidence that a DNA sequence is an inherited mutation, DNA samples from affected families are studied via linkage or association studies. If the DNA mutation associates with affected members in the pedigree significantly more frequently than dictated by chance, the DNA sequence studied is probably at or near the chromosomal locus responsible for the phenotype.

Although most mendelian disorders that include in the phenotype an increased risk for brain tumors are relatively rare in the general population, there are some exceptions. Neurofibromatosis is the most common of the syndromes with brain tumors as a phenotypic feature. Germline mutations in the neurofibromatosis type 1 (NF1) gene result in a syndrome characterized by café au lait spots on the skin, fibromatous tumors, and gliomas and meningiomas. Approximately 1 person in 4000 is affected with NF1 mutations. About half the cases are spontaneous and the other half are hereditary. In approximately 10% or less of these patients, a malignancy will develop from one of their multiple benign tumors.

The mismatch repair (MMR) cancer syndrome is an important syndrome that involves several different types of cancers, including some GBMs, meningiomas, and medulloblastomas.^8,11,12 It has an autosomal dominant pattern of inheritance, as do all the syndromes listed in Table 102-1. There are several DNA repair enzymes that are necessary for correct DNA MMR in a normal cell. In MMR cancer syndrome, an inactivating mutation or deletion of the MLH1, MSH2, MSH6, or PMS genes reduces the ability of the cell to identify and correctly repair DNA mismatches that occur during DNA replication. Therefore, a mutation in any of these genes can lead to an increased risk for brain and other cancers. The mutations in these “mutator genes” increase the cell’s mutation rate and accelerates the acquisition of mutations elsewhere in the genome, thus leading to a significantly increased risk for the development of cancer at an early age.

Adenomatous polyposis coli (APC) syndrome increases the risk for colon cancer and brain tumors, similar to MMR syndrome. However, APC syndrome, or familial adenomatous polyposis syndrome, is molecularly and phenotypically distinct from MMR cancer syndrome. This disorder increases the risk for medulloblastomas rather than gliomas and is due to a mutation in the APC gene. Turcot’s syndrome was originally applied to any syndrome with colon cancer and brain tumor but, when used now, applies only to MMR cancer syndrome.

Genetic inheritance of cancer risk is of more than scientific interest. Inherited patterns help researchers locate the genes that can initiate brain tumors, and identifying a familial risk for cancer is also good clinical practice. A new patient with familial clustering of brain tumors or other types of tumors will often want to know whether there is a gene responsible and what the risk is for their family members. Identifying the gene responsible gives the patient options for genetic counseling and family planning, as well as provides the opportunity for early diagnosis and treatment of other family members carrying the disease-causing gene allele. In some cases, identifying a cancer syndrome in a family can be lifesaving for a family member.

Tumor Suppressors, Oncogenes, and Mutator Genes

Cancer-causing gene mutations can be classified by how the mutation contributes to tumor formation. Gain-of-function mutations are changes that either enhance the normal gene function or add a new function. Gain-of-function mutations are activating and change a proto-oncogene to an oncogene. For oncogenes, only one copy of the gene needs to be mutated for activation. Activation of an oncogene then contributes functionally to tumor progression, in concert with other gene mutations. Oncogenes are typically activated either by genomic amplification of the region containing the gene or by small mutations that alter the protein sequence. Occasionally, splicing-related mutations that add or delete gene exons can activate an oncogene.

Loss-of-function mutations inactivate tumor suppressors. A tumor suppressor is a gene that normally keeps malignant progression in check. Mutations that produce a new stop codon (truncating mutation), delete all or part of the gene, disrupt the gene promoter, alter splicing, or change an amino acid or acids rendering the protein nonfunctional are all common ways to inactivate a tumor suppressor.

An important third class of cancer-causing genes is the mutator gene. This is a gene that when mutated, increases the mutation rate in the cell’s DNA. This elevated mutation rate then leads to critical mutations in oncogenes and tumor suppressors, although many new mutations will occur throughout the entire genome. Most of the mutator genes identified to date are DNA repair enzymes; when these genes are lost because of inactivating mutations, the cell subsequently loses its ability to edit and repair errors that occur during DNA replication. The resulting accelerated mutation rate then leaves the cell primed to activate oncogenes and inactivate tumor suppressors at an accelerated rate. This class of cancer gene includes the genes responsible for MMR syndrome, mentioned earlier. In GBMs, about 5% of tumors will have MMR instability, thus indicating a mutation in one of the DNA MMR enzymes.

Different Types of DNA Mutations and Alterations

DNA damage is the basis of cancer and can occur in several different ways. Mutations are changes in the DNA sequence that are acquired during the life of the cell, although in common usage they are sometimes thought of as inherited disease-causing sequences. However, mutations can occur in somatic cells and be inherited in germline cells. Mutations are more accurately and broadly defined as any DNA change that increases risk for a disease or directly promotes disease formation.

DNA alterations can range from single base pair changes all the way to entire chromosome gains or losses, as well as any size change in between. There does not seem to be any restriction to the DNA sequence changes that can occur during the development of cancer. In addition to DNA sequence changes, it also appears that epigenetic changes can contribute to tumor progression by altering gene expression without altering the nucleotide sequence.¹⁷

Point mutations are single base pair changes. In the coding regions of genes, a point mutation that alters the three-letter genetic code in such a way that the amino acid is changed is referred to as a nonsynonymous change. A synonymous (silent) mutation is one that does not alter the amino acid at that position. Normally, these silent changes are thought to be nonfunctional, but there may be hidden regulatory sequences within the coding region that can cause a functional change. Point mutations may also change an amino acid to a stop codon and, along with other mutations that induce early termination of protein translation, are referred to as truncating mutations. Point mutations and other changes can also alter gene regulatory regions in the gene or at regulatory regions distant from the gene. Other common small mutations can alter gene splicing, alter transcript levels, or form new proteins.

Insertion and deletion of one or more bases can have the same effect as point mutations. New amino acids can be added or deleted to a protein and thereby either activate a new function or delete the normal function.

If both copies of a gene are deleted, it is known as a homozygous deletion. Homozygous deletions are sometimes observed in a cancer genome and are frequently a signal to the researcher that a tumor suppressor gene was located in the lost region of the genome. More frequently, only one copy of a gene is deleted in its entirety, and this occurs in regions of loss of heterozygosity (LOH). LOH is often the first hit in inactivating a tumor suppressor in sporadic cancers, with the second inactivating smaller mutation occurring in a tumor suppressor gene within the region of LOH.

When the number of alleles is increased substantially beyond the normal two copies, it is known as genomic or gene amplification. The presence of increased copy numbers of a gene as a result of genomic amplification is a reliable indication that an oncogene is located in the amplified region. Genes such as Mouse Double Minute 2 homolog (MDM2) and epidermal growth factor receptor (EGFR), for example, are frequently amplified in GBMs. Either the normal gene can be found to be amplified (and simply increases its normal function to pathologic levels), or a mutated oncogene can be found in the amplified region. If a normal gene sequence is genomically amplified, it is still regarded as an oncogene if it results in increased expression of the gene’s protein and the increased levels promote tumor growth. Mutated genes can also arise in genomically amplified regions, and the mutation might occur either before or after the amplification.

A common large DNA sequence change that can be observed by cytogenetic techniques is chromosomal translocation. The functional consequences of chromosomal translocation can be many. In the region of the translocated gene, expression can be greatly altered. The obvious culprit of a translocation is the fusion gene, where a new oncogene is formed from the parts of two genes, similar to the prototype BCR/ABL fusion gene formed by the Philadelphia chromosome translocation in leukemia.

Some of the basic concepts of cancer genes, such as gene amplification, oncogenes, and tumor suppressors, are diagrammed in Figure 102-1B.

The Spectrum of Mutations that Underlie Brain Cancers

What mutations contribute to the formation of brain tumors? The answer, in short, is that many mutations and other DNA changes concurrently promote tumor formation and growth. The number and combinations of mutations that can be used by the cancer cell to escape the normal control mechanisms are far larger and more complex than originally envisioned by researchers. However, some general conclusions can be drawn. Although the different genes that can be mutated to form a tumor are numerous for a particular tumor type, these mutations cluster into a smaller number of pathways of basic cancer mechanisms. The common pathways include those that control cell cycle, growth factor signaling, and the p53 checkpoint.

In a recent comprehensive genomic analysis of 22 GBM genomes, in excess of 20,000 genes were sequenced for acquired mutations and assayed for changes in copy number.⁷ More than 40 genes had acquired genetic changes that occurred at a statistically significant rate and very likely contributed to the development of GBM. The mutations occurred in many various combinations. In GBMs the mutations include not only the frequently mutated and well-known genes such as EGFR and p53 but also a complex mixture of lesser known genes that are mutated in only a small percentage of cancers. This mixture of a few highly mutated genes plus many more low-frequency mutations was also observed in colon and breast cancers,^18,19 thus suggesting that it is a common theme in malignancies and will be observed in other brain cancers besides GBM.

This complex pattern of individual mutations suggests that at a genetic level there are many ways to alter the pathways leading to this most common brain cancer. Fortunately, because these molecular pathways are at least partly characterized and understood, it gives rise to the hope that brain tumors can be better classified by their molecular changes and that critical points in the pathways might serve as useful therapeutic targets.

In the following sections, attention is devoted to describing the mutations that give rise to the common types of brain tumors. It is important to note that there are few, if any mutations specific to a particular cancer. Many genes are observed to be mutated in varying degrees in multiple cancer types, and few genes are mutated at 100% prevalence in a particular histopathologic class of tumors. An individual cancer will probably have more than 20 functional mutations that give rise to its progression from a normal cell. The various combinations of genetic mutations that can give rise to a particular cancer probably make each tumor unique simply by virtue of the resulting DNA sequence with its various mutations.

Glioblastomas

GBM, the most common malignant brain tumor, is also the most studied brain tumor, and consequently its genetic basis is better described than that of other brain cancers.

GBMs have two major classes based on clinical manifestations: primary (or de novo) and secondary (or progressive). Primary GBMs first occur as a grade IV tumor, whereas secondary GBMs initially occur as a grade II or grade III astrocytoma.

About 5% to 10% of GBMs arise from grade II or III tumors.^20,21 The secondary GBMs as a group, not surprisingly, have a different spectrum of mutations. Secondary GBMs have a higher p53 mutation frequency (65% versus 28%), lower frequency of PTEN mutations (4% versus 25%), and lower rate of EGFR amplification (8% versus 36%) than do primary GBMs.²² It also appears that the types of p53 mutations differ between primary and secondary GBMs,²⁰ thus suggesting a different mode of mutagenesis.

Recent oncogenomic studies have resulted in the GBM genome being one of the best characterized for mutations.^7,23 One study sequenced the coding region of 20,661 genes and the copy number changes in 22 glioblastoma genomes followed by sequencing of an additional 83 samples for mutated genes.⁷ The integrated analysis revealed that in GBMs an average of 60 acquired genomic changes could be observed per cancer genome. Not all these mutations are causative, and some fraction probably consists of passenger mutations that do not contribute to the development of cancer. However, an individual tumor still has many alterations that contribute to tumor development, in contrast to previous predictions.

In the screening of 20,661 genes, isocitrate dehydrogenase 1 (IDH1) was found to be mutated in 11% of tumors.⁷ These mutations were found in nearly all secondary GBMs (those that progress from lower grades) and not frequent in primary (de novo) GBMs.⁷ This suggests that IDH1 and related mutations may help define the secondary GBM class. It also appears that this mutation is very highly associated with lower grade astrocytomas and is maintained in relapsed higher grade tumors. Thus, it appears that some of the pathways that lead to GBM differ and that secondary GBM can be characterized by IDH1 mutations, in addition to having a lower EGFR and higher p53 mutation rate.^6,20

As a group, there were 41 genes in the 22 GBM genomes sequenced that were statistically significant for their mutation frequency (CAN-genes)—strong evidence that at least 41 different genes can be mutated in GBMs that contribute in some way to progression of glioblastoma.⁷ Many of these genes are well-known oncogenes and tumor suppressors such as p53, but the function of the other genes with regard to cancer is still unknown. A list of the more commonly mutated genes in GBM, assembled from two major genomic studies,^7,23 is presented in Table 102-2. It is clear that not only does each tumor have many mutations but that there is also a wide variety of genes that can be mutated, thus giving each tumor a spectrum of sequence alterations that is probably unique.

Despite the complexity caused by the numerous possible combinations of the many different genes that are mutated in cancer, the mutations can be clustered into functional groups. For example, genes that might serve the same function when mutated in cancer all activate a particular signal transduction pathway. One of the most commonly mutated pathways in cancer, the p53 checkpoint, can be disrupted either by inactivation via p53 mutation or by amplification of MDM2 or MDM4, which have a dominant-negative effect. In all, half of GBMs have a p53 pathway gene altered.⁷ A sketch of the three major pathways that contribute to GBM formation is shown in Figure 102-2.

FIGURE 102-2 Some of the major pathways activated in glioblastoma multiforme (GBM). More than 40 genes have been observed to be mutated in GBM at a significant frequency, and many of these genes tend to cluster into well-known cancer pathways. The most frequently activated oncogene in GBM, epidermal growth factor receptor (EGFR), is activated by amplification or mutation, or both, in more than 40% of GBMs. It is a growth factor receptor that can activate several downstream pathways, such as those controlled by the PTEN tumor suppressor or the phosphatidylinosine-3′-kinase oncogenes (PIK3R1, PIK3CA). Other growth factor receptors in the tyrosine kinase family, such as platelet-derived growth factor receptor (PDGFR) and multiple endocrine tumors (METs), are also activated, but at a lower mutation rate than EGFR. Mutations in the growth factor signaling pathway not only stimulate growth but also prevent apoptosis and stimulate tumor cell invasion. As with the majority of cancers, the p53 checkpoint and cell cycle control are also disrupted. Cell cycle control is most commonly disrupted in GBM by loss of the CDKN2A (CDK4 inhibitor p16-INK4) tumor suppressor. Oncogenes are shown in green and tumor suppressors in pink. PIP₂, phosphatidylinositol-4,5 bisphosphate; PIP₃, phosphatidylinositol-3,4,5-trisphosphate.

Another important group of genes in the development of GBM, as well as many other cancers, is the phosphatidylinositol-3′-kinases (PI3Ks).²⁴ In half of GBMs there is either a PI3K-activating mutation, deletion/mutation of the negative regulator PTEN,^25,26 or a mutated insulin-mediated activator of PI3K kinase, insulin substrate receptor 1 (IRS1). It is also possible that mutations in their receptor tyrosine kinases, in particular amplifications of EGFR, contribute to the signaling in this pathway. Together, these genes broadly form a growth factor signaling pathway that consists of the receptor tyrosine kinases and downstream signaling through AKT, mitogen-activated kinases, and other transducers.

Of the important genes shown in Figure 102-2, one deserves special attention as the most frequently mutated oncogene known in GBM. EGFR is genomically amplified²⁷ in about 40% of tumors,²⁸ and an additional number of GBMs have an activating point mutation.^7,23 In all, nearly half of GBMs have an EGFR gene activated by a genomic change that occurs during development of the tumor.

After genomic amplification the EGFR gene may undergo further rearrangement.^29,30 The most frequently observed genomic deletion in EGFR, present in two thirds of GBMs with EGFR amplification, is loss of the internal exons in the gene that correspond to the coding sequence for amino acids 6 to 273.³¹ This most common rearrangement is referred to as the EGFR type 3 rearrangement, or EGFRvIII.³² Removal of the deleted domain constitutively activates the tyrosine kinase activity of the mutant protein.^33,34 Re-expression of the mutant protein in GBM cells increases the ability of the cells to invade through a simulated extracellular matrix.³⁵

As a highly expressed oncogene on the cell surface of about half of GBMs, EGFR is in theory a good molecular target. Efforts to treat GBM patients with EGFR inhibitors in a single-agent regimen, however, have thus far not achieved reproducible increases in survival in clinical trials.^36,37 Some of the reasons for this lack of success might include the inability to achieve sufficient intratumoral concentrations of the inhibitors or the ability of the cancer cell to signal growth through parallel pathways and thus overcome the effects of the EGFR inhibitors.

In addition to the receptor tyrosine kinase signaling pathway that is activated in GBM and many other cancers, there are several other well-studied pathways in GBM that have been defined functionally for which the causative mutations have been identified. The retinoblastoma (RB) pathway, for example, is activated in more than two thirds of GBMs by mutations in just three genes: RB1, CDK4, and CDKN2A (CDK4 inhibitor p16-INK4). RB1 and modulators of its function have the primary task of regulating entry of the cell into the cell cycle. Without the RB1 checkpoint, cell proliferation can proceed in cancer without this control.

Although MMR gene mutations can be inherited, they can also occur sporadically to initiate tumor formation or promote resistance to chemotherapy.^38–41 Because MMR mutations induce many nonfunctional as well as functional mutations, it is hard to precisely estimate the mutation rate of these genes in GBM, but about 5% of GBMs have microsatellite instability, a sign of a functional MMR mutation. It appears that MMR mutations are more common in relapsing patients, and this instability might be selected for as a means of more rapidly acquiring drug resistance.^41,42

Low-Grade Astrocytomas and Oligodendrogliomas

The lower grade astrocytomas have not been evaluated for mutations as extensively as GBMs have, but there are some molecular similarities among GBM and grade II and grade III astrocytomas. For example, p53 mutations are found in grade II astrocytomas, similar to GBMs, although at a higher rate.^7,23,43–45 It appears that of all the astrocytomas, gemistocytic astrocytomas (grade II) exhibit the highest rate of p53 mutation, with 88% having a mutation, followed next by fibrillary astrocytomas, with 53% mutated.²² Secondary GBMs have a higher p53 mutation rate than primary GBMs do.²² Yet progression of grade II astrocytomas to higher grade tumors does not seem to be dependent on p53 mutations inasmuch as grade II astrocytomas with and without p53 mutations progress at similar frequencies.^43,44 Because p53 mutations are more prevalent in grade II astrocytomas, those that progress to GBMs result in secondary GBMs with a higher rate of p53 mutation. The mutations in p53 can be an initiating event for grade II astrocytomas, as evidenced by Li-Fraumeni syndrome, in which an inherited p53 mutation gives rise to higher risk for glioma along with many other cancers.

In contrast to high-grade astrocytomas, grade I pilocytic astrocytomas have a very low frequency of p53 mutations. The low frequency at which the mutations might occur has been reported differently in the literature and probably has yet to be accurately established.^46–48 Overall, very little of the molecular basis of these mostly benign tumors is known. Despite sharing histologic features with other astrocytomas, there is little reported molecular similarity.

More is known about the molecular basis of oligodendrogliomas, and some of the chromosomal changes in this tumor help define this class. Oligodendrogliomas are now best defined at a molecular level by LOH on chromosomal arms 1p and 19q and a relative paucity of p53 mutations.^22,49,50 The loss of 1p/19q in oligodendrogliomas very accurately predicts sensitivity to chemotherapy.^50,51 It is one of the few examples of brain cancers in which a clinically useful prognostic marker has been identified.

Medulloblastomas

Table 102-3 summarizes the important known mutations for medulloblastomas. Because medulloblastoma is the most common malignant pediatric brain tumor, there have been numerous molecular studies on this cancer. Many important pathways and genes have been implicated in medulloblastoma, although systematic sequencing of the medulloblastoma coding genome has not been performed, as with GBM. Therefore, it is likely that there will be many new genes and pathways implicated in medulloblastoma in the future.

The pathway best studied thus far in medulloblastoma is the Hedgehog developmental signaling pathway. This pathway was first implicated in medulloblastoma by the discovery of Patched 1 (PTCH1) mutations.^52,58–60 PTCH1 mutation in one allele was first identified as the hereditary basis of basal cell nevus syndrome (Gorlin’s syndrome) and, with both alleles mutated, as the basis of basal cell carcinoma.⁶¹ Patients with Gorlin’s syndrome also have a significant increase in the incidence of medulloblastoma, and such mutations were also found in these tumors. Sporadic medulloblastomas also have PTCH1 mutations, but this is limited mainly to the 15% or so medulloblastomas of the desmoplastic variant.^52,58–60 By studying other genes in this developmental signaling pathway, additional mutations in desmoplastic medulloblastomas were discovered, for example, in the SUFU gene.^54,55 It appears that Patched pathway mutations help define the desmoplastic variant of medulloblastomas.

Table 102-3 also shows a low frequency of medulloblastoma mutations in other well-studied cancer-related pathways. Mutations in APC and β-catenin implicate the important colon cancer–associated APC pathway in this brain tumor,⁵³ although these mutations are found in only a small percentage of patients and are partially tied to inherited forms of this tumor. The PI3K pathway, a pathway activated in many cancers, appears to be activated in at least 5% of medulloblastomas.⁵⁷ However, the molecular basis of the majority of medulloblastomas is still unknown. It is likely to be a mixture of the few genes listed in Table 102-3, plus a large group of genes yet to be implicated.

Perspectives in Brain Tumor Genomics and Genetics

The fields of cancer genetics and cancer genomics have for several decades identified many of the molecular changes that give rise to cancer. There has been a tremendous increase in the rate of discovery of mutated genes in brain and other cancers. This acceleration is due in part to advances in automated sequencing technology and completion of the human genome sequence.^62–64 Starting with efforts such as linkage analysis, with which it took many years to identify a cancer-causing gene, technology has advanced to the point where tens of thousands of genes can be simultaneously evaluated in a cancer genome, thus enabling researchers to identify the more complete pattern of many acquired mutations in one project. This approach has already been applied to GBMs, but as the cost of these technologies decreases, one can expect to see it applied to other brain tumors. Now that we have or soon will have better knowledge of the mutations that give rise to brain cancers, attention can better be focused on understanding how these mutated pathways give rise to cancer.

There are several important reasons to understand the mutational basis of brain cancer. First, it helps us answer the question of why someone gets a brain cancer, a question frequently faced by patients and patient families. Second, the mutations and genomic alterations that occur in the more than 120 different types of brain tumor⁶ are starting to help us better classify these tumors for better diagnostic and prognostic purposes. Third, understanding of how the tumor differs at a molecular level from normal cells has helped us design successful new treatment strategies in other cancers. It is possible that this understanding will lead to better treatment options for brain tumors.