Você está na página 1de 6

1

MAGNETIC RESONANCE IMAGING OF THE BRAIN


Ferenc A. Jolesz, M.D.

To understand MRI's role in the diagnosis of central nervous system diseases we will investigate two major disease categories: malignant brain tumors and white matter lesions. The advantages and limitations of the MRI method to define and characterize these pathologies will be discussed and current advances in the neuroradiologic approach in demonstrating the tissue changes will be shown.

THE RELATIONSHIP OF TUMOR STRUCTURE TO IMAGE INTERPRETATION Description of gross anatomic abnormality on images usually implies welldefined tumor mass and associated abnormalities, such as edema and mass effects. However, the ability to discriminate those features which determine the spatial distribution and malignant potential of tumor cells is critical to the correct interpretation of images of gliomas. There have been attempts to define the structural organization of gliomas based upon patterns of spatial dissemination of tumor cells within the brain parenchyma: according to the Daumas-Duport classification, in some tumors (type 1 or well-marginated solid tumors) neoplastic cells are seen only in the core and no isolated tumor cells are found at the periphery of the mass. Type 2 is characterized by a high central density of tumor cells with isolated neoplastic cells at the periphery. Type 3 structure consists purely of isolated tumor cells infiltrating the otherwise intact normal brain. In the last case, cellular density may only minimally exceed that of the normal white matter. Various degrees of spatial dissemination may occur within the same tumor mass. This is especially evident in the case of glioblastoma multiforme in which there is marked heterogeneity and variability of the geometry, extent, and structural characteristics of the tumor. Tumor cell density generally has a centrifugally diminishing distribution, high in the core (equivalent to type 1 structure) and very low at the periphery (equivalent to type 3). On the other hand, low-grade

astrocytomas are often characterized through-out by regions equivalent to type 2 or 3 tumors. It is important to mention that this structural classification relies solely on the distribution of cellular elements and is therefore independent of the malignancy grade (which itself can vary within a tumor). For example, low-grade benign infiltrating astrocytomas and malignant anaplastic astrocytomas may present with the same pattern of cell dissemination, and glioblastoma multiforme can reveal all three of the pattern in various parts of the tumor. However, the application of the Daumas-Duport classification to imaging is limited. The tumor definition on images depends upon the signal intensity changes within the smallest image elements, the voxels. Signal intensity within a voxel is determined by the actual tumor cell density and other factors such as the presence of edema. This partial voluming of tumor, edema, and normal tissue prevents an exact definition of tumor-edema boundary (which may not, in fact, be present as tumor and edema may coexist). Tumor boundary definition based on contrast enhancement may also be misleading since low tumor cell density beyond the enhancing margin is possible. However, the application of a classification system based on the spatial dissemination of tumor cells within the brain parenchyma is necessary to understand the appearance of tumors on images. Instead of defining distinct types of tumor, we define low density, moderate density, and high density of tumor cells combined with varying degrees of spatial dissemination. This model is applicable to any tumor or to any portion of a heterogeneous neoplastic mass, e.g., glioblastoma multiforme, and is more directly applicable to imaging findings. The model describes the entire spectrum of dissemination from solid tumor mass to scattered tumor cells within normal brain tissue and includes the changes related to tumor invasion such as BBB damage and consequent contrast enhancement and edema. From the point of view of image interpretation, it is helpful to relate some of the diagnostic features of images to the degree of tumor cell dissemination within the brain. As the ratio of tumor density changes there is a concurrent change in various structural and functional properties which influence the lesion's appearance on the images. Different imaging techniques vary in sensitivity and specificity to detect alterations in cellular density, BBB, functioning tumor vascularity, and peritumoral edema. Diagnostically important lesion detection features, such as hypodensity on non-enhanced CT

or high signal intensity (T2-prolongation) on MR images, can correspond to a wide range of pathologic processes. Cystic tumor mass without intervening parenchyma, brain tissue infiltrated by isolated tumor cells, or edematous brain with or without neoplastic elements may all exhibit the same signal characteristics. T2-prolongation on MR images may demonstrate isolated tumor cell infiltration beyond the extent shown by hypodensities on CT, but cannot specifically distinguish tumorous from edematous brain, particularly in cases in which tumor cells infiltrate edematous brain parenchyma. Contrast enhancement, a sign of the functional or structural disruption of the BBB, generally defines a margin corresponding to a high density of tumor cells, but usually does not include regions in which the brain is infiltrated with isolated tumor cells. Contrast enhancement can therefore correctly outline tumors and distinguish them from edema if they have a low degree of spatial dissemination, but fails to do so with neoplasms in which tumor cells are extensively disseminated. Due to these limitations, the usefulness of images for therapeutic planning depends upon the specific clinical situation. For a surgical intervention (including radiosurgery and brachytherapy) which requires the definition of a target volume, imaging methods are sufficient when the mass is sharply marginated and consists of densely packed tumor cells. In cases of more infiltrative and disseminating tumors, imaging modalities such as CT or MRI may or may not define the entire affected tissue volume (including neoplasma and edema) without necessarily succeeding in estimating the specific contributions of various components. While functional imaging technique (SPECT, PET, MR spectroscopy) provide an additional dimension to imaging by depicting metabolic changes within the imaged volume, all current clinical imaging methods ultimately are limited to some degree in their ability to extract tissue-specific information. In addition, the sensitivity of all these methods is limited by volume averaging. Recent attempts to image tumor cells specifically using their immunogenic or membrane properties (receptorbinding labeled ligands) may eventually yield more sensitive and specific clinical methods.

CHARACTERIZATION OF WHITE MATTER PATHOLOGY

Magnetic resonance imaging (MRI) has been shown to be an extremely sensitive but not necessarily specific method for detecting lesions in the cerebral white matter (WM). Its high sensitivity explains the successful application of MRI for the diagnosis of the most common demyelinating disease, multiple sclerosis (MS), or for the detection of peritumoral edema, but is has also led to an unprecedented "diagnostic epidemic" of WM disease from various known and unknown etiologies. The whole concept of WM pathology is now under serious reinvestigation. Progress in characterizing and classifying WM pathology by applying MRI is severely inhibited by the lack of experimental correlation between MRI and conventional histopathology, and by the lack of a solid pathophysiological concept. When clinical and laboratory tests confirm the diagnosis of MS, MRI's poor specificity is less of a shortcoming. It is clear, however, that MRI is currently unable to further classify lesions as active or inactive, predict the course of disease, or confirm remissions and exacerbation; nor can intrinsic components of the MS plaques, such as edema, inflammation, or demyelination be distinguished. Nonetheless, MR characterization of WM lesions is MS has become essential to assess disease progress and to follow therapeutic regimens. Both discrete deep WM lesions and the confluent, mostly periventricular WM lesions, can be related to the local segmental demyelination of MS, but there is no distinguishing MR signal feature which can differentiate these lesions from those seen in several other pathologic process of the WM. Periventricular high signal intensity is seen anterior to the frontal horns in normal patients and around the wall of the lateral ventricles in hydrocephalus. The most likely cause in these cases transependymal migration of cerebrospinal fluid (CSF). It is also observed in Alzheimer's disease, vascular dementia (Binswanger disease, lacunar state, multi-infarct), and normal pressure hydrocephalus. Incidental WM lesions are commonly seen on MRI examinations of asymptomatic patients. The prevalence and clinical significance of these incidental WM lesions are unknown but their incidence correlates with age and with cerebrosvascular risk factors. It is thought that they most likely represent ischemic encephalopathy, or minor WM infarctions (atrophic perivascular demyelination); they occur, however, in a significant number of younger patients which suggests more widespread and divergent etiologies. The differential diagnosis of these lesions includes other obvious

causes of WM edema, such as metastasis, stroke, trauma, and leukodystrophies. Secondary Wallerian degeneration from any brain damage, post-irradiation or post-chemotherapy changes, toxic, metabolic, or infectious leukoencephalopathies, AIDS or other viral diseases, should also be considered. There is a growing disparity between the abundance of MRI data and the paucity of pathologic information explaining WM changes. An intense WM MRI signal may bear little relationship to the structural or staining abnormalities seen on histologic section following fixation and dehydration. Prolongation on T2 may be the result of CSF penetration into normal or ischemic WM through dilated perivascular (Virchow-Robin's) spaces, or due to fluid within cystic areas of microinfarcts, or be related to demyelination. Abnormal signal from the WM can be a reflection of pathologic increase in tissue water and/or changes in the macromolecular water environment due to the destruction of hydrophobic myelin membranes. Both edema and demyelination will eventually prolong the relaxation processes. The pathophysiologic mechanisms cited in WM edema and demyelination do not necessarily explain the mechanisms which determine the MR characteristics of the lesion. A full understanding of the MR representation of these lesions will require the combination of measurements which reveal both spatial distribution of water compartments and their dynamic relations with each other.

SUGGESTED READING 1. Yuh WTC, Nguyen HD, Tali ET, et al. Delineation of gliomas with various doses of MR contrast material. AJNR 1994; 15:983 Asari S, Makabe T, Katayama S, et al. Assessment of the pathological grade of astrocytic gliomas using an MRI score. Neuroradiol 1994; 36:308 Madison MT, Hall WA, Latchaw RE, et al. Radiologic diagnosis, staging, and follow-up of adult central nervous

2.

3.

system primary malignant glioma. Radiol Clin North Am 1994; 32:183 4. Lee DH, Vellet AD, Eliasziw M, Vidito L, Ebers GC, Rice GP, Hewett L, Dunlavy S MR imaging field strength: prospective evaluation of the diagnostic accuracy of MRI for diagnosis of multiple sclerosis at 0.5 and 1.5 T. Radiology 1995; 194:257 Olson EM, Healy JF, Wong WHM, et al. MR detection of white matter disease of the brain in patients with HIV infection: fast spin-echo vs conventional spin-echo pulse sequences. AJR 1994; 162:1199 Finelli DA, Hurst GC, Amantia P Jr, Gullapali RP, Apicella A Cerebral white matter: technical development and clinical applications of effective magnetization transfer (MT) power concepts for highpower, thin-section, quantitative MT examinations. Radiology 1996; 199:219

5.

6.

Você também pode gostar