Measures of cerebral blood flow (CBF) and cerebral blood volume (CBV) within tumors using MRI ¹^–⁴ or, more recently, CT⁵^,⁶ provide quantitative correlates of what has been known since cerebral angiography as one of the mainstays for brain tumor localization and characterization, that is, that tumors frequently contain dysplastic blood vessels that exhibit blood (contrast) transit time that is markedly slower than is that within the surrounding parenchyma. In the case of many tumors, catheter angiography would demonstrate a tumor blush, indicative of an abnormally enlarged blood pool. MRI or CT perfusion data correlate reasonably well with tumor grade, with higher grade tumors exhibiting greater CBV.^1,⁷^–⁹ There is also some evidence that blood volume measurements have prognostic value, with low-grade tumors that have higher blood volumes advancing in grade more rapidly than tumors with similar histology but lower blood volumes.¹⁰

Functional Magnetic Resonance Imaging

The early 1990s saw the development of imaging techniques which could assay dynamic changes in blood flow and blood oxygen extraction as a proxy for regional brain neural activity.¹¹^,¹² These techniques became known as fMRI and were based on the observations that the concentration of deoxyhemoglobin in blood affected the nuclear magnetic resonance parameter known as T2*—with a higher concentration being associated with a shorter T2*¹³—and that with brain activation, there is an increase in regional cerebral blood flow out of proportion to the increase in oxygen extraction.¹⁴ With regional brain activation, the disproportionate increase in blood flow compared with oxygen extraction leads to a higher concentration of oxyhemoglobin in the venous efflux and thus a reduced concentration of venous deoxyhemoglobin. The reduced concentration of deoxyhemoglobin and concomitant prolongation of T2* leads to a regional task-related increase in signal.

Fundamentally, all fMRI experiments rely on comparing signal intensities measured with and without presumed brain activation and mapping those regions where signal changes correlate with the temporal profile of the activation paradigm. In most cases, the experiments are set up in a so-called block design, in which epochs or blocks of task (e.g., finger movement or speech-related tasks) are alternated with epochs of rest, or at least activity not directly related to the task.

Soon after its introduction, investigators proposed fMRI as a tool in preneurosurgical planning ¹⁵ and established that its ability to localize primary sensorimotor cortex compared reasonably well with electrophysiologic techniques.¹⁶^–¹⁹ In the neurosurgical setting, the earliest and probably still most common applications of fMRI have been in localizing the hand-arm representation in the primary sensorimotor regions, although with the development of a broad array of paradigms, interrogation of language-related brain regions has become virtually as common (Fig. 109-2). fMRI has been shown useful in demonstrating the degree to which regions of eloquent cortex have been displaced by tumors or have undergone reorganization because of regional cortical dysplasia or dysfunction.²⁰^,²¹ In 1999, Lee and colleagues at the Mayo Clinic published a summary of their experience using fMRI in the setting of presurgical evaluation of tumor and seizure patients.²¹ They found that they were able to use fMRI to identify the primary sensorimotor region in 70% of patients; their success rate was close to 90% when they considered only the patients they evaluated using more recently developed data acquisition techniques.

FIGURE 109-2 Functional MRI results obtained with cued bilateral finger movements (A) and verb generation (B) in a 22-year-old patient with a left parietal giant cell ependymoma (white arrow in left upper image).

Like any technique, fMRI does not always work; the patients who are being evaluated frequently move more than healthy volunteers, which will almost certainly have an adverse effect on the quality of the data,²² as will reduced vasoreactivity in the vicinity of tumors or vascular malformations. As techniques have improved over time, especially for rapid data acquisition and for patient motion correction, the applicability of fMRI as a neurosurgical planning tool has increased.

Diffusion Tensor Imaging and Tractography

Although fMRI to estimate the proximity of functional cortex to a tumor has been available since the mid-1990s, techniques for estimating the proximity of major white matter tracts have developed much more recently, and as of this writing (mid-2008) remain under active development. These techniques are based on the biophysical observation that water molecules within axons move in a relatively unconstrained manner longitudinally along the length of the axon, whereas those molecules can move only a short distance radially before they encounter either microtubules or the axonal membrane.²³^–²⁶ Thus, if one were to measure and plot within any given imaging voxel the ability of water molecules to move in a relatively unconstrained way, one would end up with a blimp or cigar-shaped plot, oriented roughly along the axes of the axons that run through that voxel. This plot represents graphically what is known mathematically as the diffusion tensor. Conceptually, by assessing the directionality of water diffusivity in each small imaging voxel and then linking the measures of maximal diffusivity end to end, one can begin to discern the trajectories of white matter tracts as they run through the brain (Fig. 109-3, and Fig. 109-4 for a clinical example). Ongoing technical developments include improvements in tracking white matter bundles through edema surrounding tumors (and even through tumors) and in separating white matter tracts that closely approximate one another, or that cross one another within a group of voxels.²⁷^–³¹

FIGURE 109-3 Diffusion tensor imaging (DTI) and tractography. A, Spherical representation of the diffusion tensor that one would find without the directionality imposed by a white matter tract. B, Elongated diffusion tensor representation, with the long axis of the ellipse corresponding to the dominant direction of the axons (thin tubes) within the voxel. C, Approximate trajectory of a white matter tract (red line) estimated by linking end to end a series of elongated diffusion tensor representations such as depicted in B.

FIGURE 109-4 Tractography in a patient with a right cerebral hemisphere glioblastoma multiforme. A, Axial T2-weighted images showing cross sections of the tractographic representations of the left (pink) and right (blue) corticospinal tracts at the level of the mass. Note the medial and anterior displacement of the right tract due to mass effect from the tumor. B, Reconstruction of the tracts, showing medial displacement of that on the right due to the mass.

Often fMRI and DTI-tractography results can be used complementarily because in the vicinity of brain tumors or vascular malformations, the coupling between neural activity and the reactive changes in blood flow on which fMRI signal changes depend is often impaired. In such cases, the fMRI data may be of reduced utility, whereas the tractographic data may still be quite robust (Fig. 109-5). Conversely, extensive edema may render the tractographic data suboptimal, whereas the fMRI signal changes in the overlying cortex remain relatively unaffected.

FIGURE 109-5 Functional magnetic resonance imaging (fMRI) (A) and DTI-based tractography (B) in a patient with a left occipital arteriovenous malformation. The patient’s visual fields are completely intact bilaterally, yet multiple visual fMRI tasks resulted in markedly asymmetric signal change, presumably because of impaired vasoreactivity in the vicinity of the lesion. The tractographic data, however, show slight lateral displacement of the optic radiations. These data aided in surgical planning for lesion resection.

The additions of advanced neuroimaging techniques notwithstanding, the first decision a neuroradiologist must make when evaluating an intracranial mass is whether the lesion is intraparenchymal or extra-axial. Extra-axial masses frequently exhibit at least some of the following characteristics:

• An obtuse margin with the brain parenchyma

• Buckling of the cortex

• Contrast enhancement of a dural tail

• Infiltration of or reaction by the overlying bone

• Dilation of the subarachnoid space at the margins of the lesion

• Contrast enhancement of cranial nerves or the leptomeninges

• Relative absence of vasogenic edema

Although each of these findings may also be identified with lesions that are intracranial (especially infiltration of or reaction by the overlying bone and relative absence of vasogenic edema), a combination of these features would suggest an extra-axial process.

Among the extra-axial processes, five are most common: meningioma, schwannoma, lymphoma, metastases, and granulomatous diseases (most notably sarcoidosis). Schwannomas occur in stereotypical locations associated with cranial nerves and rarely have dural tails of enhancing tissue arising from the margins of the lesion. In the case of dural or epidural metastases and dural lymphoma, the clinical picture usually suggests the diagnosis because of prior histories of a primary malignancy, systemic lymphoma, or HIV disease (although intraparenchymal central nervous system [CNS] lymphoma is far more common than dural-based lymphoma in the AIDS setting). Sarcoidosis may affect the pia, dura, or parenchyma and will commonly have pulmonary manifestations. It may elicit more of a parenchymal inflammatory and edematous reaction than the other diagnoses. Other granulomatous diseases such as tuberculosis or fungal infection are differentiated based on systemic symptoms. In the end, meningiomas still predominate in the extra-axial compartment, and unless there are unusual features as described previously, this is the most likely diagnosis to consider. Particularly with a dural tail and fine calcification or bony reaction, meningioma is the top choice.

Features that suggest that masses are intra-axial include the following:

• Containment of all the margins of the lesions within the brain parenchyma

• No sharp margins with the dural surface

• Thickening of the cortex rather than buckling

• Absence of enhancement of the overlying leptomeninges or infiltration of the overlying bone

• Extensive vasogenic edema

When lesions grow through the margins of the dura, as in some aggressive meningiomas or cases of glioblastoma multiforme (GBM), the analysis of these lesions becomes much more difficult.

These caveats aside, the presence of necrosis, edema, and calcification and the location are the main factors that suggest a specific histologic classification. Most higher grade astrocytomas, primitive neuroectodermal tumors (PNETs), and lymphomas enhance. If a lesion does not enhance, one is likely to be dealing with a low-grade astrocytoma (pilocytic astrocytomas excluded), ganglioglioma, or subependymoma. Absence of enhancement virtually excludes metastases, hemangioblastomas, and GBM.

Necrosis is often evident in high-grade astrocytomas and usually implies a GBM. Although lymphomas in the non-AIDS population rarely show necrosis, such necrosis is much more often seen in AIDS-related CNS lymphomas. Metastases often demonstrate central necrosis. Often one must distinguish among cyst formation (seen frequently in hemangioblastoma, pilocytic astrocytoma, desmoplastic infantile ganglioglioma [DIG], dysembryoplastic neuroectodermal tumor [DNET], and ganglioglioma) and necrosis. The latter is usually much more irregular and elicits more edema in the surrounding tissue.

Edema may be absent with low-grade astrocytomas, gangliogliomas, DIGs, ependymomas, hemangioblastomas, some PNETs, and some DNETs. The lesions with the greatest degree of edema are the lymphomas, glioblastomas, and metastases. A lesion such as gliomatosis cerebri may infiltrate without evoking edema. Most lesions that evoke edema also produce mass effect unless they are merely infiltrating the cortex like gangliogliomas and DNETs.

Calcification occurs frequently with oligodendrogliomas, neurocytomas, and craniopharyngiomas. Nonetheless, by virtue of their higher incidence, astrocytomas still represent the most common calcified tumor. The calcification of an oligodendroglioma tends to be coarser than the more stippled calcification of astrocytomas. The metastases that calcify include mucinous adenocarcinomas, osteosarcomas, and chondrosarcomas.

Finally one must assess location. A lesion confined to or based predominantly in the gray matter likely represents a ganglioglioma, DNET, cortical dysplasia, or pleomorphic xanthoastrocytoma. Lesions that cross the corpus callosum are usually high-grade astrocytomas or lymphoma. Metastases do not as a rule cross the corpus callosum. Hemangioblastomas usually occur in the cerebellum, often bordering on the pia. Neurocytomas are found most commonly along the septum pellucidum. Oligodendrogliomas favor the temporal lobes, as do DNETs, DIGs, and pleomorphic xanthoastrocytoma. Clearly, the differential diagnosis of a pineal region mass differs from one in the suprasellar region (although germinomas and meningiomas may break that rule) or one that is in the ventricle. Location is a critical piece of the analysis.

This initial review offers insight into the framework that is used to diagnose most intracranial masses, with the understanding that radiology rarely preempts a histologic specimen, particularly for the intra-axial masses.

Extra-Axial Masses

Meningioma

By far, the most common extra-axial neoplasm to affect the intracranial compartment is the meningioma. This tumor is characterized by hyperdensity relative to normal brain parenchyma on CT, isointensity on T1-weighted images relative to gray matter, and slight hyperintensity on T2-weighted images (Fig. 109-6). It should be noted, however, that there is a wide variety of signal intensity characteristics associated with meningiomas, and in fact, some reports suggest that syncytial, transitional, and angioimmunoblastic meningiomas may have differing signal intensity characteristics depending on their internal histology. Nearly all meningiomas show strong contrast enhancement on either CT or MRI. Meningiomas may show stippled or confluent calcification. If the meningioma is extensively calcified, enhancement may be less evident.

FIGURE 109-6 Planum sphenoidale meningioma. A, Much of the mass on this unenhanced sagittal T1-weighted image demonstrates signal that is close to isointense to brain parenchyma (arrow) but is sharply demarcated from the immediately overlying rostrum and genu of the corpus callosum. B, On the axial T2-weighted image, the mass is again seen to displace (arrows) rather than invade or arise from the brain parenchyma. Except for the center of the mass, much of it exhibits T2-weighted signal close to that of the surrounding brain. C, Coronal postgadolinium T1-weighted image shows the marked enhancement and a small dural tail (arrow) projecting leftward from the base of the mass.

Meningiomas also have a characteristic “dural tail,” which represents contrast enhancement extending along the margins of the tumor affecting the pachymeninges (Fig. 109-7). Some histologic studies have suggested that the entirety of the dural tail represents meningioma tumor, whereas others have suggested that this may represent reactive change adjacent to neoplasm.³²

FIGURE 109-7 Convexity meningioma. This lesion, which has extended extracranially (black arrow), as well as intracranially, demonstrates a “dural tail” (white arrows) that extends down over the right hemispheric convexity.

Bony lysis or sclerosis may be seen in about 30% of patients who have meningiomas. Viewing the CT scans using bone windows may be useful in detecting the increased density and overgrowth of the bone and the lytic areas adjacent to the meningioma. Although these areas usually represent a reaction by osteoblasts to the tumor, in some cases the meningioma may permeate the bone.

The spectroscopic features of meningioma on proton spectroscopy include absence of NAA and elevations of choline peaks. The height of the NAA and creatine peaks may be markedly reduced. Alanine has been suggested to be a specific marker for meningiomas, but its presence is variable.³³^–³⁵ Although this may help differentiate a meningioma from an intra-axial mass, many other extra-axial masses (see later) have a similar spectroscopic signature.

Meningiomas may deviate from the characteristic benign appearance in many different ways and on occasion demonstrate necrosis, fatty degeneration, cystic areas, infiltration into the brain, infiltration into the bone, and marked vasogenic edema.³⁶ The presence of vasogenic edema associated with meningiomas has been correlated with the lesion size as well as the degree of parasitization of dural venous structures.

On dynamic imaging with contrast agents, the meningiomas show slow uptake of the contrast agent in a continuous fashion followed by a lengthy period of stable enhancement and a delayed clearance of the contrast agent. MRI-based estimates of CBV has been shown to help differentiate between meningioma and dural-based metastatic deposits, with the meningiomas typically showing substantially greater CBV.³⁷ Angiographically, dural vessels generally supply the lesions. This usually means that the branches of the external carotid artery, including the middle meningeal artery or stylomastoid branches of the occipital artery, supply the mass. Nonetheless, branches of the tentorial artery from petrous carotid meningeal branches may be responsible for the primary supply of tentorial meningiomas. Similarly, meningiomas around the cavernous sinus may have direct carotid branches supplying the lesion. Meningiomas at the foramen magnum may receive blood supply from branches of the vertebral artery or posterior inferior cerebellar artery.

Schwannoma

The next most common extra-axial mass is the schwannoma, dominated by those that occur in and around the internal auditory canal. This lesion characteristically resides in the cerebellopontine angle or in the internal auditory canal, arising most often from the vestibular branches of the 8th cranial nerve (Fig. 109-8). Its imaging characteristics are similar to those of a meningioma from the standpoint of being relatively isointense to gray matter, but the absence of a dural tail usually helps to distinguish these two lesions. Vestibular schwannomas, however, may show cystic degeneration as well as hemorrhage and occasionally cause edema in adjacent brain tissue. As opposed to meningiomas, it would be highly unusual for an acoustic schwannoma to cause bony lysis or bony sclerosis. In particular, if a cerebellopontine angle lesion shows enhancing tumor entering the internal auditory canal, one would favor vestibular schwannoma over meningioma.