The original classification scheme proposed by Bailey and Cushing in the 1920s serves as the foundation for the histological categorization of all brain tumors currently proposed by the World Health Organization (WHO). Basically, the WHO classification scheme recognizes seven major categories based on the cell of origin (Table 5.1). These include tumors of neuroepithelial cells, tumors of the nerve sheath (composed of Schwann cells and fibroblasts), tumors of the meninges (composed of meningothelial, mesenchymal, and melanocytic tumors), tumors of lymphoproliferative cells, tumors of germ cell origin, tumors of the sella, and metastatic disease. Each of these cells of origin gives rise to a particular tumor type. The tumors of neuroepithelial origin comprise the largest group and include tumors of glial origin (usually astrocytic tumors, oligodendroglial tumors, ependymal tumors, choroid plexus tumors), tumors of nonglial origin (e.g., ganglioglioma, central neurocytoma, and others), tumors of pineal origin (e.g., pineocytoma and pineoblastoma), and embryonal tumors (e.g., medulloblastoma and primitive neuroectodermal tumor).

The cell of origin directly impacts on tumor nomenclature. If the cellular composition is primarily astrocytes, then the tumor is called an astrocytoma. If the majority of the cells are oligodendroglial, then it is termed an oligodendroglioma, and so on. The brain itself is predominantly composed of neuroepithelial cells and hence the most common tumor type is derived from this cell line. Although the neuron is the most common cell type overall in the brain, mature neurons do not divide and thus cannot produce neoplastic growth. Therefore, most (40% to 50%) tumors of the brain itself are gliomas. Since they arise from the brain parenchyma itself, these tumors are virtually always “intra-axial” in location.

Since most nonglial tumors do not arise from neuroepithelial tissue in the brain parenchyma, they are overwhelmingly located outside of the brain proper, that is, along the coverings of the brain or within the ventricular system. Tumors arising from these “outside” locations are therefore referred to as “extra-axial” tumors. This concept of intra-axial and extra-axial is critical to the correct interpretation of cross-sectional CNS imaging studies. In general, the histologic composition of these tumors is directly linked with the location of the tumor.

The clinical presentation of a CNS neoplasm is almost always related to increased intracranial pressure, seizure activity, or a focal neurologic deficit.

Subfalcine Herniation. The falx is a very tough fibrous structure that is very resistant to any sort of displacement. When a mass is located in certain key locations or is of sufficient size, portions of the brain itself may be pushed across the midline or through dural openings. This displacement of the brain is called herniation and many types have been defined according to the brain structure that is most affected. Subfalcine (or cingulate) herniation is the most common type of herniation and occurs when the cingulate gyrus is displaced under the margin of the interhemispheric falx. Even if the cingulate gyrus is not displaced underneath the falx, significant general midline shift is considered to be present if the shift is 3 mm or greater.

Uncal and Central Herniation. The uncus represents the hooked extremity of the parahippocampal formation of the medial temporal lobe. Uncal herniation often compromises the many tracts running through the brainstem as well as cranial nerves, particularly the oculomotor (III) nerve causing an ipsilateral pupillary dilation (or “blown pupil”). On imaging studies, effacement of the ambient cistern and contralateral hydrocephalus are the hallmarks of uncal herniation. Central herniation is the result of either downward or upward displacement of the brainstem through the tentorial insura. It most commonly results from bilateral or midline supratentorial masses that cause complete obliteration of the cisternal spaces. Elongation of the brain stem in the anteroposterior axis and narrowing in the transverse axis is seen on axial images.

Hydrocephalus. The mass effect of an intracranial neoplasm may be sufficient by itself to produce increased intracranial pressure or hydrocephalus secondary to obstruction of the flow of cerebrospinal fluid (CSF) as it circulates through the ventricles and into the subarachnoid space. The increased pressure is associated with a classic clinical triad of headaches, nausea and vomiting, and papilledema (caused by partial obstruction of the venous outflow from the optic nerve). These features may occur at any time during the course of the brain tumor. In addition, altered mental status (particularly with bifrontal lobe tumors) or alterations in equilibrium
(commonly seen in cerebellar or eighth cranial nerve tumors) may be present. Intracranial neoplasms usually present with an indolent course marked by progressive headache and focal neurologic deficit, but may also present abruptly.

The detection of an intracranial abnormality on any imaging study should immediately provoke the following three questions:

Mass? By far the most important question to ask is: “Is it a mass?” It is important to consider that abnormal attenuation on computed tomography (CT) or signal intensity on magnetic resonance (MR) imaging does not necessarily equate to a “mass,” which, by definition, must have mass effect. In other words, it must displace normal brain structures. Many diseases may produce mass effect and therefore qualify as a “mass.” However, all tumors by their very nature should have mass effect. The mass effect from a very small tumor may be beyond the limits of detection on imaging studies but this is an infrequent event. Whenever a mass is encountered on an imaging study, a neoplasm is a prime consideration in the differential diagnosis.

Distinguishing an early infarct from a neoplasm may be problematic on CT but is straightforward on MR imaging studies, especially with diffusion-weighted imaging (DWI). If these studies are not available, a follow-up conventional imaging study (preferably with MR imaging) in 3 weeks’ time may be helpful. Virtually all infarcts will be smaller in size by 3 weeks after clinical presentation. If the lesion is the same size or larger at 3 weeks, a neoplasm should be favored. Also, as detailed in Chapter 4, a subacute infarct will often show signs of subtle hemorrhage. Obviously, as the treatment of tumor and infarct are dramatically different, the distinction between a tumor and an infarct is critical for appropriate clinical management of the patient.

Intra-axial or Extra-axial? Once the presence of a mass has been determined, the next important question to ask is: “Is the mass intra-axial or extra-axial?” As presented earlier, an intra-axial mass is a mass that is of the brain itself (i.e., it arises from the brain parenchyma). An extra-axial mass refers to everything outside the brain (i.e., arachnoid, meninges, dural sinuses, skull, etc.). The ventricular system is also considered extra-axial. Determining the intra-axial or extra-axial location of a suspected neoplasm is crucial to formulating an appropriate differential diagnosis. Extra-axial lesions are characterized by “white matter buckling” or inward compression of the white matter (often with thinning of the fronds of the white matter) and maintenance of the gray matter–white matter interface (Fig. 5.1). In contradistinction, an intra-axial mass expands the white matter, thickens its fronds, and blurs the gray matter–white matter interface. However, white matter buckling is not foolproof in differentiating extra-axial from intra-axial lesions. Where extensive white matter edema is present, no buckling of the white matter may occur. Therefore, while the “white matter buckling” sign is helpful when present, its absence does not necessarily indicate that a lesion is intra-axial.

Tumor Margin? A third question often posed is: “Where’s the tumor margin?” The histologic examination of a typical brain tumor actually provides the answer. On microscopic analysis, every glioma and practically every intra-axial neoplasm lacks a capsule and therefore it is possible for neoplastic cells to migrate far from the apparent center of the tumor. The consequence is that there is no distinct margin for an intra-axial neoplasm. Therefore, knowing the margin is not possible on microscopy, it certainly is not possible by cross-sectional imaging. Treatment is typically directed to the entire region of abnormal hyperintensity on T2-weighted imaging (T2WI), not just the region described by enhancement on the T1-weighted postcontrast sequence. MRS is valuable in identifying areas of spread in regions with normal T2 signal.

Trying to make a histologic diagnosis from an MR or CT scan is fraught with hazard. However, it is possible to render an intelligent analysis of the mass, including assessment of signal intensities and enhancement characteristics of the mass, and to present an accurate differential diagnosis whenever one encounters a mass suspected of being a CNS tumor on an imaging study.

Imaging evaluation of intracranial neoplasms is best conducted by MR, which is far superior to CT because of its multiplanar capability, increased contrast resolution, and lack of ionizing radiation. CT is superior to MR in the assessment of calcification, although the use of gradient-recalled echo (GRE) or susceptibility-weighted (SWI) sequences increases the sensitivity of the latter technique to calcification. CT is invaluable
for the evaluation of bony abnormalities, such as erosion of the skull base.

MR. A basic MR evaluation of a patient suspected of having an intracranial neoplasm includes sagittal and axial T1-weighted sequences, an axial fluid-attenuated inversion recovery (FLAIR) sequence, an axial T2-weighted sequence, DWI, and postcontrast axial and coronal T1-weighted sequences. FLAIR imaging excels at demonstrating abnormal signal intensity involving the periventricular and peripheral regions but also is associated with more artifact in the posterior fossa. The unenhanced T1-weighted sequences allow distinction between inherent T1 shortening, such as in hemorrhage, and true contrast enhancement. For temporal lobe and midline lesions, the coronal plane usually provides the best delineation of the tumor. Postcontrast sagittal MR is often best for midline masses and may facilitate radiation therapy planning. Performance of all of these sequences can be easily completed within 45 minutes on 1.5 T MR units. In addition to improved contrast resolution and signal-to-noise resolution, 3 T MR scanning is being used with increased frequency and allows acquisition of volumetric sequences that may shorten the scanning. Perfusion imaging is increasingly used to assess for cerebral blood volume and other vascular parameters associated with brain tumors. Implementing gradient-moment nulling (flow-compensating) techniques facilitate evaluation of the posterior fossa by decreasing phase artifact generated by the dural sinuses. (Fig. 5.2). MR spectroscopy and metabolic imaging may also be valuable in many cases.

The cross-sectional imaging appearance of CNS tumors varies with their cellular composition and the presence or absence of hemorrhage and calcification. On CT, intra-axial neoplasms will typically appear as hypodense masses with a variable amount of surrounding white matter edema, the area of which roughly correlates with the biologic behavior of the tumor. On MR, the mass is usually dark on T1 (T1 prolongation) and bright on T2 (T2 prolongation) with variable surrounding vasogenic edema. The presence of calcification within the tumor usually produces marked hypointensity on T1WI and T2WI. Occasionally, because of the surface area of the crystals producing T1 shortening, calcification may appear bright on T1WI.

Nontumoral Hemorrhage. The appearance of intracranial parenchymal hemorrhage usually depends on the age of the blood. In hyperacute (less than 6 hours) hemorrhage, the predominant oxyhemoglobin will produce T1 and T2 prolongation (dark on T1 and bright on T2). When the hemorrhage has been present for 6 to 24 hours, the effect of deoxyhemoglobin predominates and the lesion has mild T1 prolongation (dark on T1WI) and moderate T2 shortening (darker on T2WI). After 3 to 4 days, methemoglobin begins to predominate, first being intracellular, producing T1 and T2 shortening (bright on T1WI and dark on T2WI) and then, as the red blood cells begin to lyse, becoming extracellular where the lesion has T1 shortening and T2 prolongation (bright on both T1WI and T2WI). In chronic hemorrhages (older than 10 to 14 days), hemosiderin appears, producing a rim of extreme T2 shortening. This peripheral markedly hypointense rim occurs because of migrating macrophages, which carry the hemosiderin to the periphery of the hemorrhage. On CT, acute hemorrhage (less than 1 week old) has increased attenuation (hyperdensity) compared to the normal brain tissue. By 1 to 3 weeks after the hemorrhage, the signal becomes isodense compared to the normal brain parenchyma. After 3 weeks, the focus of hemorrhage is hypodense to brain parenchyma, simulating the attenuation of CSF. This evolution of blood breakdown products is illustrated in detail in Chapter 4.

Tumoral Hemorrhage. The appearance of intratumoral hemorrhage reflects the heterogeneous nature of the tumor and is quite different compared to benign parenchymal hemorrhage. Intratumoral hemorrhage is often intermittent, producing a heterogeneous mixture of the various blood breakdown products just described. In addition, hemorrhage may occur in cystic or necrotic portions of the tumor, creating blood–blood or fluid–blood levels. Debris from the necrotic mass will also contribute to this heterogeneous mixture. Normal deoxyhemoglobin evolution is delayed such that it will persist for longer than the usual 3 to 4 days after hemorrhage. The typical hemosiderin ring does not form with intratumoral hemorrhage, probably due to interference with the migration of the macrophages by viable tumor at the margins. In cases where there is confusion as to the nature of an intracranial hemorrhage, the presence of a nonhemorrhagic mass adjacent to the hemorrhage, the persistence of T2 prolongation (most likely representing edema or tumor itself), and mass effect all suggest intratumoral hemorrhage instead of a simple parenchymal
hematoma. Gadolinium administration is often helpful in these cases, as benign hematomas should not have as significant enhancing rim as those from tumors.

Hemorrhagic Neoplasms. Because of their high vascularity, certain neoplasms are noted for their propensity to hemorrhage. Choriocarcinoma among primary tumors and metastases from melanoma, thyroid carcinoma, and renal carcinoma show this characteristic. In the setting of multiple hemorrhagic lesions within the brain, these tumors should be considered. Multiple cryptic arteriovenous malformations, either occurring de novo or secondary to radiation therapy, can have a similar appearance with the exception of surrounding vasogenic edema.

T1 Shortening. Besides hemorrhage, two other entities may produce focal T1 shortening on MR scans. Fat within lipomas or dermoids produces marked T1 shortening and intermediate signal on T2WI following the signal intensity of subcutaneous fat. The presence of chemical shift artifact on T2WI associated with such a lesion helps to confirm the presence of fat. Melanin, as seen in melanotic melanoma, also follows the same signal intensities as fat on T1WI and T2WI.

Hyperdense Neoplasms. Tumors of high cellular density, usually those with small cells such as lymphoma, pineoblastoma, neuroblastoma, or medulloblastoma, are usually hyperdense compared to brain tissue on CT. In addition, metastases from melanoma, lung carcinoma, colon carcinoma, and breast carcinoma may be hyperdense. On MR, these same tumors are typically hypointense on T2WI with the appearance presumably being related to a high nucleus-to-cytoplasm ratio of the tumor cells, which produces less free water and thus less T2 prolongation. On occasion, iso- or hyperintensity may be seen because of heterogeneity of the tumor matrix.

Enhancement. Contrast enhancement, whether from iodinated contrast agents used in CT or paramagnetic gadolinium agents used in MR, occurs based on one primary factor: breakdown of the blood–brain barrier. Unlike nonneural endothelium, the endothelium of the cerebral capillaries allows the passage of only small molecules through their tight junctions and narrow intercellular gaps. The macromolecules that make up contrast agents are too large to pass this barrier under normal circumstances. When the blood–brain barrier breaks down, contrast is able to leak across the barrier and abnormal contrast enhancement is seen. Many pathologic states, including intra-axial tumors (either primary or metastatic), inflammatory diseases, subacute infarcts, postoperative gliosis, and radiation necrosis among others, may be associated with this event. Some tumors, particularly low-grade neoplasms, will not show enhancement presumably because they form new capillaries that are quite similar to the native cerebral capillaries with the blood–brain barrier left intact. More biologically aggressive high-grade neoplasms tend to have fenestrated capillaries that allow the passage of contrast media and consequently show image enhancement. However, the fact that a lesion enhances means that there is a breakdown of the blood–brain barrier, and the presence or absence of enhancement cannot be used to categorically state that a lesion is low-grade or high-grade (Fig. 5.3). In addition, some specialized areas of the brain, such as the choroid plexus, pituitary and pineal glands, tuber cinereum, and area postrema, have no blood–brain barrier and will normally enhance after administration of a contrast agent.

In the evaluation of a postoperative brain tumor patient, timing is of the essence. It is recognized that postoperative granulation tissue develops within 72 hours following surgery and enhances after administration of contrast. Once formed, this tissue may persist for weeks to months. Since most malignant brain tumors have at least some enhancement, the presence of enhancement on a study performed during this 72-hour window is generally regarded as being related to residual tumor rather than granulation tissue.

Safety. Postoperative neurosurgical patients are often not ideal candidates for scanning in an MR unit and proper monitoring of vital signs to assure their safety is of paramount importance. If the appropriate monitoring and life-support equipment (e.g., shielded pulse oximeter and oxygen) and personnel are not available to safely perform an MR study, a monitored contrast-enhanced CT should be substituted.

Many malignant tumors will be treated by a combination of chemotherapy and radiation therapy following surgical debulking. Typical radiation doses are in the range of 5000 to
5400 rad, most often delivered in fractionated doses (about 180 rad each visit) over several weeks time. As a consequence, radiation injury in the white matter occurs in two forms: diffuse white matter injury and radiation necrosis.

Despite the widespread involvement seen in diffuse white matter injury, most patients do not have any neurologic deficits. A “geographic distribution” of white matter hyperintensity on T2WI conforming to the selected radiation ports is typical for diffuse white matter radiation injury and should not be misinterpreted as vasogenic edema from the tumor. Virtually all patients following whole-brain or large-volume radiation will demonstrate this pattern of involvement 6 months or more after the therapy is completed. Affected areas do not enhance on postcontrast imaging. Distinction from a brain tumor is seldom a problem in the setting of diffuse white matter injury.

In contrast, the much less commonly seen radiation necrosis is virtually indistinguishable from that of recurrent tumor on CT or conventional MR images as both usually have mass effect and enhancement. Clinically, patients with radiation necrosis often present with focal neurologic deficits. However, MRS, DWI, perfusion imaging, and metabolic imaging (PET or SPECT-thallium studies) are useful in making this distinction. While radiation necrosis may have an elevated lactate peak on MRS secondary to necrosis, the elevated choline peak and depressed N-acetyl aspartate (NAA) peaks expected with tumors are usually absent (Fig. 5.4). Necrotic areas typically show restricted water diffusion (hyperintensity on DWI and corresponding hypointensity on apparent diffusion coefficient (ADC) map images). In contrast, the overwhelmingly majority of brain tumors do not cause restricted water diffusion and will not be hyperintense on DWI. Similarly, perfusion imaging may distinguish between radiation necrosis and tumor, as some tumors (particularly those of higher grade) show increased perfusion while there is absence of increased perfusion in radiation necrosis. With metabolic imaging, radiation necrosis shows normal to decreased metabolic activity while recurrent tumor is usually increased, especially if the original tumor was high-grade (e.g., WHO grade III or IV; see “Astrocytomas” section). Distinguishing between recurrent tumor and radiation necrosis using metabolic imaging is less accurate if the original brain tumor was a WHO grade I or II tumor.

It is difficult, in many circumstances, to suggest a specific histologic diagnosis based on the imaging characteristics alone. However, taking into account other factors such as the location of these tumors (intra-axial, extra-axial, intraventricular, sellar region, pineal region) and clinical information (age, gender, endocrinologic data, etc.), the differential diagnosis can be limited to just a few possibilities and sometimes a single most likely entity. Some intracranial tumors have a definite predilection for one gender (see Table 5.2).