| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Departments of Diagnostic Radiology and Organ Imaging (Y.L.C., A.D.K., C.M.) and Clinical Oncology (S.F.L., P.H.K.C.), Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong, People's Republic of China. Received June 23, 1998; revision requested September 3; revision received March 3, 1999; accepted July 1. Address reprint requests to Y.L.C. (e-mail: yl190chan@cuhk.edu.hk).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: This was a prospective study involving 34 patients (age range, 37–72 years) with known radiation injury to the temporal lobes from radiation therapy administered 2–10 years previously for nasopharyngeal carcinoma. MR imaging was performed with T2-weighted gradient- and spin-echo, gradient-recalled echo, T1-weighted spin-echo, fluid-attenuated inversion-recovery, and T1-weighted postcontrast spin-echo sequences.
RESULTS: Radiation injury was present in 57 of the 68 temporal lobes. The white matter lesions in radiation-induced injury were predominantly hyperintense on T2-weighted images, but in 37 (65%) of the 57 lobes, foci with heterogeneous signal intensity consistent with necrosis were detected. In the 57 involved lobes, gray matter lesions were detected in 50 (88%); blood-brain barrier disruption based on parenchymal contrast enhancement, in 51 (89%); and hemosiderin deposits, in 30 (53%). There was a significant correlation between white matter necrosis, gray matter lesions, and blood-brain barrier disruption, all of which were located mainly in the inferior temporal lobes that received the highest radiation dose.
CONCLUSION: The lesion components of radiation-induced injury to the temporal lobes at MR imaging were more varied than have been previously described. In addition to the classic white matter lesions, gray matter lesions, blood-brain barrier disruption, and hemosiderin deposition also were frequently seen.
Index terms: Brain, MR, 134.121411, 134.121412, 134.121413, 134.12143 • Brain, necrosis, 10.47, 134.47 • Gadolinium • Magnetic resonance (MR), pulse sequences, 134.121411, 134.121412, 134.121413, 134.12143 • Radiations, injurious effects, complications of therapeutic radiology, 10.47, 134.47
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Nasopharyngeal carcinoma is one of the most common malignant tumors that affects the southern Chinese population. The standard treatment for nasopharyngeal carcinoma is radiation therapy, which involves irradiation of the inferior part of the temporal lobes. Radiation-induced temporal lobe necrosis is one of the most serious complications of radiation therapy for the treatment of nasopharyngeal carcinoma; it accounts for 65% of the irradiation-related deaths from nasopharyngeal carcinoma in Hong Kong (8–10), and the 5-year probability of surviving temporal lobe necrosis, with or without treatment, in Hong Kong has been reported to be 59% (11). Therefore, this type of necrosis is a clinically important entity, even though it has an incidence of only 5% in 10 years (12). Furthermore, uncomplicated nasopharyngeal carcinoma is extracranial, and radiation injury to the temporal lobes occurs in an originally intact brain unaffected by tumor or tumor-associated edema. Thus, the changes in the temporal lobes induced by radiation therapy represent radiation injury to the native brain because the brain abnormalities detected are caused solely by radiation.
A more detailed morphologic assessment is needed to understand the disease process of late radiation injury to the brain and its subsequent staging and management. Therefore, the objective of the present study was to perform a more thorough examination of the morphologic characteristics of late radiation injury to the temporal lobes of the brain at MR imaging.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The nasopharynx and adjacent regions received a radiation dose of 66.0–71.2 Gy (the target volume dose). A three-port arrangement, which consisted of one anterior and two lateral opposing 6-MV photon beams, was used during either a part of or the entire course of the radiation therapy. The superior margin of the ports was 0–1 cm above the plane of the planum sphenoidale and was adjusted according to the extent of the tumor. Eye shields were placed in the anterior radiation port. The lower margin of the eye shields was on a plane that passed through the inferior orbital margin and the floor of the pituitary fossa. Therefore, the most inferior portion of the temporal lobes below this plane was incidentally irradiated by all the beams and thus received the target volume dose. The part of the temporal lobes above this plane but below the superior margin of the ports received approximately two-thirds of the target volume dose because of the shielding of the anterior beam by the eye shields. The inferomedial portion of this part of the temporal lobe received the full target volume dose, because the medial margin of the eye shields was 1.75 cm from the midline and their inferior margin was convex.
Six patients received a second course of radiation therapy for recurrence of tumor. Patients with clinical, endoscopic, or imaging evidence of recurrent tumor, and clinical or laboratory evidence of central nervous system infection at the time of the study were excluded.
MR imaging was performed by using a 1.5-T MR imaging unit (Gyroscan ACS NT; Philips Medical Systems, Best, the Netherlands) with a standard head coil. The following imaging sequences were performed:
1. Sagittal and coronal T2-weighted gradient- and spin-echo (GRASE): repetition time (TR) msec/echo time (TE) msec, 4,550/90; turbo factor, seven; echo planar imaging factor, three; rectangular fields of view, 70% (coronal section) and 80% (sagittal section); matrix, 256 x 256; field of view, 25 cm; number of signals acquired, two; section thicknesses, 4 mm (coronal) and 5 mm (sagittal) with 10% intersection gap; and use of an inferior regional saturation band.
2. Coronal fast-field echo (ie, gradient-recalled echo [GRE]): 300/30; flip angle, 30°; rectangular field of view, 70%; matrix, 256 x 256; field of view, 25 cm; number of signals acquired, two; section thickness, 4 mm with 10% intersection gap; and use of an inferior regional saturation band.
3. Axial T1-weighted spin-echo: 500/15; rectangular field of view, 70%; matrix, 256 x 256; field of view, 25 cm; number of signals acquired, two; section thickness, 5 mm with 10% intersection gap; and use of an inferior regional saturation band.
4. Axial fluid-attenuated inversion-recovery: TR/TE/inversion time msec, 8,000/120/2,400; rectangular field of view, 70%; matrix, 256 x 256; field of view, 25 cm; number of signals acquired, one; section thickness, 5 mm with 20% intersection gap; and use of an inferior regional saturation band.
5. Axial T1-weighted postcontrast spin-echo: 500/15; rectangular field of view, 70%; matrix, 256 x 256; field of view, 25 cm; number of signals acquired, two; section thickness, 5 mm with 10% intersection gap; use of an inferior regional saturation band; and obtained after an intravenous bolus injection of 20 mL of gadodiamide (Omniscan; Nycomed Amersham, Princeton, NJ) in a concentration of 0.5 mmol/mL. In addition, coronal T1-weighted spin-echo images were obtained in the last 18 patients examined.
The diagnosis of the lesion components at MR imaging was based on the following criteria: A white matter lesion was defined as any abnormal signal intensity change in the white matter on T2-weighted GRASE images. A homogeneously hyperintense white matter change was defined as edema or demyelination. When the white matter lesion was not homogeneously hyperintense but contained internal hypointense foci that could not be attributed solely to hemosiderin deposits on GRE images, it was defined as white matter necrosis, because such an appearance could not be accounted for merely by edema or demyelination. A gray matter lesion was defined as a hyperintense change in the cortex on T2-weighted GRASE images.
Hemosiderin deposition was diagnosed when GRE images showed angular, streaklike, curvilinear, or lacelike foci that were more hypointense than normal white matter. For nodular hypointense foci to be diagnosed as hemosiderin deposits, they had to be markedly hypointense—to the extent that they could be differentiated from necrosis. Those foci that were not clearly demarcated from the hypointense bone of the skull base were excluded to eliminate the possibility of them mimicking susceptibility artifact from the skull base. The diagnosis of blood-brain barrier disruption was based on the detection of a contrast material–enhanced focus in the temporal lobe brain parenchyma after contrast material injection. Mass effect was graded as absent or nonsubstantial, moderate, or severe. Severe mass effect was diagnosed when there was a midline shift. Moderate mass effect was diagnosed when the temporal lobe was enlarged, with elevation and obliteration of the Sylvian fissure but without a midline shift.
The MR images were assessed independently by two neuroradiologists (Y.L.C., A.D.K.), and the findings of whether the described lesion components were present were evaluated for strength of agreement by using Cohen 
statistics (SPSS 6.0; SPSS, Chicago, Ill) to determine fair ( < 0.60 but 
0.40), good ( < 0.75 but 0.60), or excellent ( 0.75) agreement (13). The final interpretation of the results was reached by consensus between the two neuroradiologists; a third investigator (S.F.L.) was available for arbitration when an agreement could not be reached. The Spearman rank correlation test was used to test the significance of the correlations between white matter necrosis, gray matter lesions, hemosiderin deposition, blood-brain barrier disruption, and mass effect. The Student t test for means was used to assess for any significant relation between the interval after radiation therapy and the detection of each lesion component. The study was approved by the ethics committee of the university, and written informed consent was obtained from all patients.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Lesion Characteristics
Interobserver agreement for the diagnosis of different lesion components was as follows: good for white matter necrosis ( = 0.68), excellent for gray matter lesions ( = 0.75), excellent for hemosiderin deposition ( = 0.85), excellent for blood-brain barrier disruption ( = 0.84), and fair for mass effect ( = 0.57).
White matter lesions.—White matter necrosis, seen as heterogeneous lesions in the form of ill-defined foci or confluent patches of mild hypointensity within hyperintense white matter lesions on T2-weighted GRASE images, was detected in 37 of the involved temporal lobes (Fig 1a). Twenty-six of these 37 temporal lobes also had hypointense areas amidst the hyperintense white matter lesions on the fluid attenuated inversion-recovery images (Fig 1b). Well-marginated homogeneous and markedly hyperintense areas consistent with frank cystic changes in the white matter were present in nine of the 37 involved temporal lobes (Fig 2). Homogeneously hyperintense lesions consistent with demyelination or edema were detected in 20 of these involved temporal lobes. These lesions involved either the deep white matter (Fig 3) or the entire white matter, including the deep and subcortical portions (Fig 1a).
|
|
|
|
|
|
|
|
|
2 cm) enhancing lesions tended to have an irregularly enhancing rim. Twenty-four of the 28 larger lesions with an irregularly enhancing rim contained internal linear and punctate foci of enhancement, which produced a mosaic configuration. Gray matter lesions characteristically showed nodular enhancement (Fig 5b). Faint rim enhancement was seen in the well-defined cystic lesions. Hemosiderin deposition.—The GRE images showed hypointense foci consistent with hemosiderin deposition in 30 involved temporal lobes. These foci were either nodular, angular, or in various linear configurations (eg, curvilinear, streaklike, or lacelike) (Fig 6a). On the T2-weighted GRASE images, which were substantially less sensitive for the detection of this component, hemosiderin deposition was detected in 16 of the 30 involved temporal lobes, and the amount and conspicuity of the deposits were much less compared with those seen on the GRE images (Fig 6).
|
|
Mass effect was seen in 17 involved temporal lobes in 12 patients. Severe mass effect (ie, with a midline shift) was detected in four patients (Figs 1, 6), three of whom had unilateral temporal lobe involvement. The fourth patient had bilateral but asymmetric involvement. Moderate mass effect was seen in 12 temporal lobes in eight patients. In four patients, there was bilateral involvement, with a similar degree of temporal lobe swelling on both sides, obliteration of the Sylvian fissures, and cerebral sulci. Four patients had asymmetric involvement; the more severely affected side showed enlargement of the temporal lobe with obliteration of the Sylvian fissures and no midline shift.
Relations between Lesion Components
Of the 51 temporal lobes with blood-brain barrier disruption, 49 had evidence of either gray matter lesions or white matter necrosis on the basis of the presence of hypointense foci in a hyperintense lesion in white matter (Fig 1a, 1c). Only two temporal lobes with blood-brain barrier disruption had no evidence of white matter necrosis or gray matter lesions. Conversely, the contrast-enhanced images of two temporal lobes with gray matter lesions or necrosis were negative for blood-brain barrier disruption. Blood-brain barrier disruption was detected on the axial contrast-enhanced images of 36 of the 37 temporal lobes with white matter necrosis. In the 18 patients in whom coronal T1-weighted postcontrast MR imaging also was performed, all of the 15 temporal lobes with white matter necrosis on GRASE images showed evidence of blood-brain barrier disruption (Fig 1a, 1c).
When the occurrence of a lesion component in the 57 involved temporal lobes was correlated with the occurrence of another lesion component, significant correlations existed among gray matter lesions, white matter necrosis, blood-brain barrier disruption, and mass effect (Table). The occurrence of hemosiderin deposition correlated significantly with the occurrence of gray matter lesions but not with the occurrence of white matter necrosis, mass effect, or blood-brain barrier disruption. There was no significant difference in the interval after radiation therapy in the patients who had a specific lesion component compared with those patients who did not. This was true with regard to white matter necrosis, gray matter lesions, blood-brain barrier disruption, hemosiderin deposition, and mass effect.
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MR imaging has been found to be more sensitive than CT for the detection of radiation-induced temporal lobe necrosis (9). MR imaging combined with T2 relaxation measurement has been proposed for early diagnosis (16–18) and noninvasive monitoring (19) of radiation-induced brain changes in animal studies. Norris et al (14) described gray and white matter lesions, cystic components, and contrast enhancement patterns on the MR images obtained in seven patients with late radiation-induced injury of the brain. With the inclusion of 34 patients and use of more MR imaging pulse sequences, in the present study we sought to investigate the frequency and characteristics of more lesion components.
The results of previous studies suggest that radiation injury to the brain involves predominantly white matter (2,6,9), with hyperintense foci demonstrated on T2-weighted MR images (2,5). These foci might become confluent (2) and result in an irregular hyperintense area on T2-weighted images and a low-signal-intensity area on T1-weighted or intermediate-weighted images (9). Pathologically, radiation-induced focal injury in white matter varies from rarefaction of myelin and reactive gliosis to coagulation necrosis and cavitation (4).
The results of the present study revealed that the extent of white matter lesions was greater than that of gray matter lesions, blood-brain barrier disruption, or hemosiderin deposition. Although it is uncommon for white matter lesions to occur alone, they may be the only abnormal finding at MR imaging. White matter lesions on T2-weighted images may be homogeneously or heterogeneously hyperintense. The homogeneously hyperintense lesions in our study showed no enhancement after contrast material administration. Extensive homogeneously hyperintense lesions, especially when they are beyond the confines of the area of irradiation, are probably due to edema. It has been shown histologically that low-attenuating white matter changes at CT are due to reactive white matter edema (8). Severe edema that spreads progressively to parts of the brain that are remote from the area of irradiation has been observed in dogs (20) that were treated with irradiation.
Demyelination also may be a component of white matter changes (4,21), especially in the less extensive lesions; histologic proof of demyelinaton in some less severely affected areas of radiation injury has been documented (8). In the present study, the heterogeneous lesions, which were considered to represent necrosis, consisted of a hyperintense lesion that contained internal low-signal-intensity areas on T2-weighted images. These lesions were demonstrated predominantly in the inferior part of the temporal lobes that received the higher dose of irradiation. These internal low-signal-intensity areas are probably liquefactions from frank necrosis, and the heterogeneity in signal intensity may be the result of a random distribution of necrotic foci of various sizes. Small, rounded low-signal-intensity areas on intermediate-weighted images within a lesion that is hyperintense on T2-weighted images have been observed, and it has been suggested that these represent central liquefaction (9).
The fact that most of the lesions with heterogeneous signal intensity in our study showed blood-brain barrier disruption also supported a diagnosis of necrosis; it has been shown that the contrast enhancement seen at CT in radiation injury is usually associated with necrosis at histologic analysis (4). Cystic lesions also were detected at MR imaging in this study. Unlike liquefactive necrosis, these lesions had a well-defined margin, typically were homogeneous in signal intensity, and showed faint rim enhancement after contrast material administration. Such lesions may represent a later stage in the evolution of white matter necrosis.
Although radiation necrosis in the human brain has been reported to be due to coagulative necrosis and mainly confined to the white matter (6,20,21), gray matter involvement was observed in an animal study (21). In the series of Norris et al (14), two of seven patients with late radiation-induced injury had gray matter involvement at MR imaging. Peterson et al (22) detected MR imaging abnormalities in both the gray and white matter in all six of the patients with radiation necrosis whom they examined. In the present study, gray matter lesions were detected in 50 (88%) of 57 temporal lobes with late radiation-induced injury. This high prevalence may have been due first to the unique setting of radiation therapy for nasopharyngeal carcinoma, in which the higher radiation dose to the inferior and inferomedial portions of the temporal lobes resulted in a higher dose to the gray matter. Second, the better delineation of the cortex with the use of coronal sections resulted in a higher sensitivity for gray matter lesion detection.
Gray matter lesions were characteristically seen as a disruption of the gray matter by hyperintense lesions on T2-weighted images. This is consistent with the reported histologic descriptions of coagulative necrosis involving the cortex and white matter (22) and of a thin ribbon of superficial cortex undercut by a ragged lamina of encephalomalacia or mummified coagulative necrosis (23). Less frequently, gray matter lesions occurred as a blurred gray-white matter junction with an ill-defined increase in signal intensity in the deep layer of the cortex on T2-weighted images. This also is consistent with previous histologic descriptions of involvement of the deep cortical layers in delayed radiation-induced injury (21).
Fibrin exudation in a lamina of hypocellularity along the gray-white matter junction has been considered by some to be a characteristic histologic finding in lesions with delayed radiation necrosis (23); this may explain the abnormalities at the gray-white matter junction on MR images. These changes at MR imaging have a pathophysiologic basis, because fibrinoid necrosis has been observed not only in the vessels in radiation-induced necrosis in white matter (8), but also in the vessels in the adjacent cortex (24). The significant correlations among the occurrences of white matter necrosis, gray matter lesions, and blood-brain barrier disruption suggest that these lesion components may share a similar pathophysiology.
The exact mechanism of blood-brain barrier disruption in radiation-induced injury is not known. The ultrastructural changes of an attenuated endothelium with fenestration, separation of endothelial cells, and increased pinocytotic vesicles have been reported to be the morphologic bases of the increased permeability of blood vessels and the subsequent disruption of the blood-brain barrier (24). Early vascular damage that progressed to blood-brain barrier disruption, edema, and necrosis in the irradiated pig brain was observed by Miot et al (18).
Multifocal punctate (<1 cm) or ring enhancement on the MR images of cranial irradiation to brain tumors has been reported (22). Intense enhancement after contrast material administration also has been seen in both gray and white matter lesions in late radiation changes in the brain (14). In the present study, blood-brain barrier disruption was detected in 51 (89%) of 57 temporal lobes with late radiation-induced injury. This high prevalence of blood-brain barrier disruption matches the findings of Norris et al (14), who reported contrast enhancement in all seven patients with late radiation-induced brain injury whom they examined. The higher detection rate at MR imaging compared with the reported detection rate of 43% at CT (8) is probably due to the higher contrast resolution of MR imaging.
In our study, the enhancement pattern varied: Small nodular enhancement, small rounded to oval rim enhancement, and irregular rim enhancement were observed. Larger lesions with evidence of white matter necrosis tended to have irregular rim enhancement with internal nodular and linear enhancing foci, which produced a mosaic appearance; this enhancement pattern suggested that these lesions formed from the coalescence of smaller lesions. Multiplication of small foci of necrosis and subsequent coalescence to give a mosaic appearance have been reported in studies of animals subjected to whole-brain irradiation (4,23).
Hemorrhage as a fatal event has been described in association with radiation-induced temporal lobe necrosis, with microscopic observation of hemosiderin granules around the hematoma (25). Patches of hemorrhage in five of 12 cases with histologic confirmation of temporal lobe necrosis also has been described (8). These hemorrhages probably resulted from radiation-induced vascular alterations—either a primary change in the vessel walls, such as acute fibrinoid necrosis, or changes secondary to telangiectasia that developed as a result of collateral drainage from thrombi or fibrin occlusion in the diseased small venules (8,25–28). Gaensler et al (28) and Pozzati et al (29) reported, respectively, the detection of hemosiderin deposition in radiation-induced telangiectasia at MR imaging and occult cerebrovascular malformations that were small, solitary, and typically not associated with necrosis or a surrounding hyperintense change.
To our knowledge, hemosiderin deposits have not been described elsewhere in association with radiation-induced necrosis on imaging studies. In the present study, hemosiderin deposits were frequently detected in temporal lobes with late radiation-induced injury on GRE MR images. These hypointense foci probably were not due to heavy calcification because calcification has not been described in previous studies of temporal lobe necrosis at CT (8,9), and calcification of the brain after radiation therapy without chemotherapy is a rare entity (30). Calvo et al (21) reported that telangiectasia preceded necrosis in all the animals with radiation-induced necrosis that they studied. In the present study, although hemosiderin deposits were frequent in late radiation injury, there was no significant correlation with white matter necrosis or blood-brain barrier disruption. This raises the suspicion that hemosiderin deposition may be a confounding event in the evolution of late radiation injury to the brain.
In summary, the lesion components of late radiation-induced injury involving the temporal lobes at MR imaging were more varied than have been described previously in the imaging literature. In addition to the classic white matter lesions, gray matter lesions, blood-brain barrier disruption, and hemosiderin deposition also were frequently demonstrated. Their relative occurrence may be related to the specific radiation treatment used in this group of patients. Nevertheless, acquaintance with these MR imaging characteristics is of value for understanding and diagnosis of late radiation injury to the brain, and these components may be cornerstones for longitudinal studies aimed at the proper staging and management of the disease process.
|
Footnotes |
|---|
Author contributions: Guarantors of integrity of entire study, Y.L.C., S.F.L.; study concepts, Y.L.C., S.F.L.; study design, Y.L.C.; definition of intellectual content, Y.L.C., S.F.L.; literature research, Y.L.C., S.F.L., A.D.K.; clinical studies, S.F.L., P.H.K.C.; data acquisition, Y.L.C., A.D.K.; data analysis, Y.L.C., A.D.K., S.F.L.; statistical analysis, Y.L.C.; manuscript preparation, editing and review, all authors.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Asao, Y. Korogi, M. Kitajima, T. Hirai, Y. Baba, K. Makino, M. Kochi, S. Morishita, and Y. Yamashita Diffusion-Weighted Imaging of Radiation-Induced Brain Injury for Differentiation from Tumor Recurrence AJNR Am. J. Neuroradiol., June 1, 2005; 26(6): 1455 - 1460. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Cha, E. A. Knopp, G. Johnson, S. G. Wetzel, A. W. Litt, and D. Zagzag Intracranial Mass Lesions: Dynamic Contrast-enhanced Susceptibility-weighted Echo-planar Perfusion MR Imaging Radiology, April 1, 2002; 223(1): 11 - 29. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
