For a long time, the classification of brain tumors has been based on histogenesis. The success of the classification proposed by Bailey and Cushing in 1926 has been associated with the hypothesis that the supposed cell of origin and the degree of maturation of tumor cells are determinant factors for prognosis. Histopathologic criteria, immunohistochemistry and, more recently, molecular genetics have improved characterization of the supposed cell of origin and have further refined the prognostic criteria. The debate about the origin of brain tumors is again of great interest after the introduction of neural stem cells and their implications for new therapeutic strategies.
The natural history of brain tumors can be divided into four main stages: initiation, progression, transformation, and diffusion. The initiation of glial tumor is not yet understood, and is the focus of intense research. When a macroscopic lesion is first identified in humans, it is already organized as a tumor. Experimental studies in rats have shown that induced tumors arise from primitive neuroepithelial (neural stem) cells of the ventricular zone (VZ), from its derivative subventricular zone (SVZ), or from a renewal pool of cells in the remnants of SVZ or subependymal layer, hippocampus, cerebellum, and first cortical layer. Neural stem cells are characterized by their self-renewal capability and multipotency.
Both neurons and glial cells derive from neural stem cells of the VZ and SVZ, and they differentiate along their respective pathways under extrinsic and intrinsic stimuli. Growth factor signaling also controls their passage from one stage to the next. Availability of markers for glia or neurons affirms that neurogenesis and gliogenesis continue in the adult mammalian brain; demonstration of the existence of neural stem cells also in the adult has changed our notion of oncogenesis dramatically. Glial tumors may arise from stem cells either in the VZ or SVZ. New hypotheses on glioma origin postulate that astrocytes and radial glia (astroglial lineage) might act as multipotent stem cells both in embryo and in adulthood. The vulnerability of stem cells to undergo neoplastic transformation depends on the interaction of several factors, including the number of replicating cells, the duration a cell population remaining in cycle, and the state of cellular differentiation.
Another important concept is that the more genetic alterations are needed for a tumor to develop the more advanced the progenitor cells’ stage of differentiation. Then the histology of a tumor may be more a reflection of the environment and time of initiation than of the cell of origin, and the former factors would determine whether a tumor ultimately becomes an astrocytoma or an oligodendroglioma. Early genetic events differ between astrocytic and oligodendroglial neoplasms, but all tumors have an initially invasive phenotype, which complicates therapy. Tumor cells are heterogeneous and variably express differentiated antigens typical of the cell of origin, but only a minority of them are self-renewing, multipotent, clonogenic, and continuously replenishing mature cells. The discovery that only a proportion of tumor cells are clonogenic when xenografted has introduced another important new concept: the existence of cancerous stem cells.
The growth and transformation of tumors can be summarized as follows. Tumors are supposed to initiate as monoclonal expansion and, due to genetic instability, they will develop genetic heterogeneity that is associated with increasing mutation rate and proliferation capacity. Once the genome becomes unstable, as the cell divides, genetic material that codes for growth promotion (i.e. oncogenes of which protein products serve to accelerate cell growth) will accumulate, whereas genetic material that codes for growth control (i.e. suppressor genes of which protein products serve as brakes on cell growth) will be lost. These events result in phenotypic heterogeneity. New clones will arise and compete with older clones. They may adapt better to the tissue microenvironment and show greater proliferation potential after having lost differentiating capacity. Anaplasia is a feature of the neoplastic tissue that has lost such capacity typical of a given cytogenetic stage, and regresses to a less differentiated stage. Various molecular genetic alterations have been linked to the following pathologic events occurring with tumor progression: proliferation and apoptosis in diffuse astrocytoma; deregulation of cell cycle in anaplastic astrocytoma; necrosis, angiogenesis, and clonal selection in glioblastoma (GBM).
Cell migration and invasion, angiogenesis, necrosis, and apoptosis are neoplastic events that have been consistently associated with poor prognosis in gliomas. New therapeutic strategies can be explored when the molecular pathways regulating these events and their phenotypic imaging characteristics become better understood.
Several molecular biology techniques are used to identify genetic aberrations in tumors. At present, the most commonly applied technique is genotype analysis with fluorescent microsatellite markers that evaluate loss of heterozygosity (LOH). Preliminary results suggest that in the near future, molecular genetic tumor analysis together with clinical, quantitative imaging and histopathologic phenotypes will allow a more accurate prediction of survival and will be of great importance in selecting and developing the appropriate therapy. At present, the use of quantitative imaging and molecular genetics for the classification of the tumor type, subtype, or grade remains a challenge because large validation studies are still lacking. It is also unknown if the correlation found between molecular genetics data and survival can be prospectively applied to each individual case at the time of diagnosis.
Growth depends on the balance between cell proliferation and cell loss, and is regulated by cell cycle time, growth fraction, tumor doubling time, necrosis, and apoptosis. Necrosis is a sudden event in which many cells are killed at the same time because of hypoxia, energy depletion, and inflammatory response. Apoptosis denotes a programmed cell death comprised of three separate complex regulatory pathways, and is considered the major cause of cell loss in gliomas and other tumors. However, apoptosis is more often associated with a high proliferation rate. The apoptotic index increases in the spectrum from low-grade diffuse (infiltrative) astrocytoma to GBM; in oligodendroglioma, the index is higher than in astrocytoma, and it also increases with anaplasia. Moreover, a high apoptotic index is found in PNETs, lymphomas, and metastases. Whether the finding of increased number of cells undergoing apoptosis might indicate tumor regression and better prognosis is still controversial. Failure of apoptosis may be responsible for tumor development and may be linked to breakage of the pathway regulated by tumor suppressor p53. In contrast, induction of apoptosis in glioma cells could be instrumental to therapies.
Diffusion and infiltration into the adjacent brain tissue are two other important properties of gliomas and make their treatment much more difficult. Motility of glioma cells and invasion are facilitated by extracellular matrix and adhesion molecules. Cell motility is associated with poor prognosis. Velocity of solid tumor expansion is linear with time and varies from 4 mm/
year in low-grade glioma (LGG) to 3mm/month in high-grade glioma (HGG). There is a cell density gradient decreasing from the center of the mass towards its periphery at the boundary with normal tissue. How far neoplastic cells can be found from the
macroscopic edge of the tumor is another very important issue. The 2 cm distance detected by computerized tomography has been considered the safety margin according to the classic report by Burger in 1988.
It is about time that new studies with more updated imaging modalities define with greater accuracy the extent of a tumor. Cell motility and invasion capability have great implications for planning of surgery and post-surgical adjunct therapy.
Cell invasion is not necessarily a consequence of cell proliferation. Examples of infiltrative but not highly proliferative gliomas are common and have been described. In solid tumors, a cellular density gradient is much more frequently found between the center and the cortex than toward the white matter. Similar gradient properties can be found in the frequency of mitoses and nuclei stained for proliferation markers, MIB1. When the tumor border is clear-cut the gradient is also steep. In infiltrated cortex, the MIB1 labeling index may be very low, because there is a dissociation between migratory and proliferation capacities. Recognition of asymmetry in tumor infiltration and proliferation also has implications for therapy planning.

The classification of tumors of the nervous system released by the World Health Organization (WHO) in 2007 is the most comprehensive to date and has received a large consensus among neuropathologists. It divides brain tumors into seven categories: tumors arising from neuroepithelial tissue, from peripheral nerves, from the meninges, lymphomas and hematopoietic neoplasms, germ cell tumors, tumors of the sellar region, and metastatic tumors.
Among tumors of neuroepithelial tissue, the gliomas are by far the most common and best studied.
Gliomas include tumors arising from neural stem cells in the VZ or SVZ or from neoplastic transformation of precursor or mature glial cells. The group of neuroepithelial tumors also includes ependymomas, neuronal and mixed neuronal–glial tumors (such as ganglioglioma), and embryonal tumors (such as medulloblastoma).
Gliomas are classified by their histologic features, according to the presumptive cell of origin, differentiation, and grade of malignancy. At the current state of knowledge, cytogenesis is more a theoretical concept than a definitive basis for tumor classification.
 

Astrocytomas are believed to originate from astrocytes, which are stellate branched cells permeating the interstitium and interacting with blood vessels and neurons. Astrocytes have indeed a very important stromal role: the extracellular matrix reacts through cell adhesion molecules and other factors with the intracellular space, modulating cell migration, differentiation, proliferation, and apoptosis. In view of the abundance of astrocytes and their multiple functions, astrocytomas make up a heterogeneous group of tumor subtypes, with different biological behavior. According to the WHO classification, glial tumors are graded on the basis of the most malignant area identified on histopathologic specimens. Fibrillary astrocytoma (WHO grade II) is characterized by increased cellularity with a monomorphic population of cells infiltrating the neuropil. Anaplastic astrocytoma (WHO grade III) is characterized by nuclear polymorphism and mitoses. The occurrence of angiogenesis and necrosis are features of GBM (grade IV). GBM can arise de novo (primary) or transform from a pre-existing LGG (secondary).
Anatomic location and patient’s age at initial presentation are also very important factors for diagnosis and prognosis. According to a recent study by Duffau and Capelle, LGGs more frequently than GBMs are found in the cortex of secondary functional areas, especially within the supplementary motor area (SMA) and insula.[15] This preferential location may be associated with a high risk of adverse postoperative sequelae, and it may be one reason that resection of LGGs remains controversial.
 

Oligodendrogliomas are moderately cellular and composed of monotonous, round, and homogeneous nuclei with a clear cytoplasm. They have a dense network of branching capillaries. It is not uncommon for oligodendrogliomas to bleed, and they may present as an intracranial hemorrhage. Additional histologic features include calcifications and mucoid/cystic degeneration. Estimates of incidence of oligodendroglioma vary enormously in different series since diagnosis
depends on use of permissive or restrictive criteria. If restricted histologic criteria are used, diagnostic signs are the honeycomb appearance of the cells and the high density and chicken-wire distribution of small vessels.
When more permissive criteria are used, the incidence of oligodendrogliomas obviously increases while that of diffuse astrocytomas decreases. Burger has masterly illustrated this diagnostic conflict. Today more than ever, the correct diagnosis of oligodendroglioma is important because effective chemotherapy has become available. Allelic LOH in the 1p and 19q chromosomes have been associated with longer survival and a favorable response to chemotherapy with procarbazine, lomustine, and vincristine (PVC). In contrast, LOH on 17p and TP53 mutations characteristic of astrocytic tumors are rare in oligodendrogliomas and practically mutually exclusive with LOH 1p and 19q. The differential diagnosis also has important prognostic implications, since the number of mitoses and nuclei positive for proliferation markers found in the two tumors may receive a different weight by the neuropathologists. In diffuse astrocytomas, mitoses are absent or very low in number and a high mitotic index indicates anaplasia. In grade II oligodendrogliomas, the number of mitoses allowed, on the other hand, is definitely higher. Therefore the same mitotic index may suggest anaplasia or not depending on the diagnosis of astrocytoma or oligodendroglioma. The apoptotic index is also much higher in oligoastrocytoma than in astrocytomas. As we will see later, this asynchronous biologic behavior between the two tumor types is a recurrent confounding factor in neuroimaging as well: the choline (Cho) signal measured with 1H-MRS and the cerebral blood volume (rCBV) measured with perfusion MR are generally higher in grade II oligodendrogliomas than in astrocytomas.
 

Mixed oligoastrocytoma (MOA) were first recognized as an entity by Cooper in 1935. The diagnosis of MOA requires recognition of two different glial components, both of which must be unequivocally neoplastic. They may be divided into biphasic (compact) and intermingled (diffuse) variants. In the former, distinct areas of both cell types are juxtaposed, while in the latter variant the two components are intimately admixed. Estimates of their incidence vary with the diagnostic criteria used and must be interpreted with caution.
Their incidence may vary from 1.8% in North American, 9.2% in Norwegian, and 10–19% in German series. On conventional MRI, MOA demonstrate no special features that would allow a reliable distinction from oligodendrogliomas. About 30–50% of MOA are characterized by LOH on 1p and 19q. About 30% carry mutations of the TP53 gene and/or LOH on 17p that are frequently found in astrocytomas. The presence of LOH 10q has also been associated with shorter overall survival in MOA. Apparently these genetic alterations are consistent throughout every individual MOA and suggest that they are monoclonal neoplasms originating from a single precursor cell rather than tumors that have developed concurrently.
 

GBM is the most malignant astrocytic tumor, composed of poorly differentiated neoplastic cells. It is the most frequent brain tumor, accounting for 12–15% of all intracranial tumors. GBM may manifest at any age, but about 80% of patients are between 45 and 80 years old. It may develop from diffuse astrocytomas, anaplastic astrocytomas, but more frequently they occur de novo after a short clinical history. Primary GBM accounts for the vast majority of cases in older people, while secondary GBM typically develops in younger patients (less than 45 years). The time to progression varies considerably, ranging from less than 1 year to more than 10 years, with a mean interval of 4–5 years.
There is increasing evidence that these two subtypes represent distinct disease entities, which evolve through different genetic pathways, and are likely to respond differently to therapy. GBM is a very heterogeneous disease and it is often multifocal. Multifocal GBM is suggestive of a more invasive and migratory tumor phenotype, a feature more common to stem cell derived cancer. According to a recent report GBM originating from the SVZ and extending into the cortex has a higher rate of developing multifocal disease, while GBM growing in the cortex has a higher rate of local recurrence. Tumor location may suggest the presumed origin of the tumor from SVZ neural stem cells in the former phenotype and from transformation of
mature glial cells in the latter. Stem cell-derived GBM may require treatment that attends to both the primary lesion and the SVZ.

Histological grading is not only a practical way of predicting the biological behavior of a tumor in an individual patient, but also an effort to define, in conjunction with other parameters, a homogeneous group of patients. Significant indicators of anaplasia in gliomas include nuclear atypia, mitotic activity, cellularity, vascular proliferation, and necrosis. These histopathological features are condensed in a three-tiered scheme. A simple and reproducible grading scheme is of paramount importance to planning therapy as well as for the interpretation of response to multiple therapeutic regimens. The validity and reproducibility of any grading system depends on the homogeneity of the lesions within each class. Grading is only one component of a combination of criteria used to predict a response to therapy and outcome. Other criteria include the patient’s age, neurological performance status, tumor location, extent of surgical resection, proliferation indices, genetic alterations, and one radiological feature: contrast enhancement. Whether 1H-MRSI and other imaging parameters will be included in this list will depend on their unique contribution to characterize neoplasms in a homogeneous group. The two biggest assets of in vivo MR imaging are the possibility to follow non-invasively the biological behavior of individual tumors and guide therapy.
For each tumor type, the combination of the above criteria contribute to an overall approximation of prognosis. The median survival for grade II diffuse astrocytoma is around 5 years; however, the range of survival is broad and unpredictable. Despite the initial low proliferative index, most of these patients die from progression to GBM. Patients with grade III anaplastic astrocytoma survive for 2–3 years; the majority of patients with GBM, in particular the elderly, have a median survival of less than 1 year. The outcome of 676 GBM patients over a 7-year period at a single institution has been reported recently: survival probabilities were 57% at 1 year, 16% at 2 years, and
7% at 3 years.
Studies of patients with grade II oligodendrogliomas reported a median survival of about 10 years. A recent series of 106 patients yielded a median survival of 16 years, probably due to earlier diagnosis following the advent of MRI. Malignant progression is not uncommon, although it is considered less frequent compared to diffuse astrocytomas. A median survival time of 6.3 years and 5- and 10-year survival rates of 58% and 32%, respectively, have been reported in a study of 60 patients with grade II MOA.