Introduction	Choose Top of pageABSTRACTIntroduction <<Materials and MethodsResultsDiscussionReferencesCITING ARTICLES

Nasopharyngeal carcinoma shows high frequencies of skull base erosion and intracranial extension [1, 2]. This infiltrative neoplasm is frequently associated with perineural infiltration [3,4,5,6]. The mainstay of treatment for nasopharyngeal carcinoma is radiation therapy. The natural history of tumor spread requires adequate radiation treatment coverage of the skull base and the middle cranial fossa. Radiation doses below 60 Gy, at conventional 2 Gy daily, are inadequate for tumor control [7]. Therefore, the effective radiation dose for the treatment of nasopharyngeal carcinoma exceeds that of the quoted tolerance limit for neural tissues, resulting in a substantial risk of radiation-induced brain damage [8,9,10]. Lee et al. [7] reported a 3% cumulative incidence of temporal lobe necrosis in a series of 4527 patients.

We report the CT and MR imaging characteristics of temporal lobe changes in patients who underwent radiation therapy for nasopharyngeal carcinoma. These findings may help radiologists differentiate temporal lobe changes from intracranial tumor recurrence.

Materials and Methods	Choose Top of pageABSTRACTIntroductionMaterials and Methods <<ResultsDiscussionReferencesCITING ARTICLES

We reviewed the film records of 1916 patients with histologically confirmed nasopharyngeal carcinoma examined over a 5-year period. The imaging studies of 47 patients (2.5%), previously treated with radiation therapy, showed temporal lobe changes. All patients were treated with a modified technique described by Ho [11] and Tsao [12]. Thirty-two patients were treated for recurrence and 15 had only a single course of treatment. We examined 38 men and nine women with a man-to-woman ratio of 4.2:1. The average age of our patients was 50 years (age range, 36-77 years).

The diagnostic criteria for temporal lobe changes include temporal lobe edema with or without foci of contrast enhancement or cavitation. No biopsy confirmation was available with the exception of one patient. All patients showed no cavernous sinus involvement or epidural mass. These criteria exclude temporal lobe changes caused by tumor infiltration (Fig. 1). The average follow-up period was 2.6 years (range, 15-113 months).

View larger version (196K)

Fig. 1. —55-year-old man with tumor recurrence showing intracerebral invasion. Coronal enhanced T1-weighted spin-echo MR image shows tumor recurrence (asterisk) associated with large extradural mass and invasion of right temporal lobe (arrow). Intracranial tumor recurrence is usually located in extradural space but may invade brain in more advanced cases.

The film reports of 1916 patients showed 47 patients with temporal changes. The imaging studies of these patients were reviewed by two radiologists. Imaging findings of temporal lobe changes were independently analyzed according to study technique. Thirty-four patients underwent CT (55 examinations) and 26 patients underwent MR imaging (32 examinations). Thirteen patients underwent CT and MR imaging. The CT and MR images of these 13 patients were separated and independently analyzed. Temporal lobe lesions were documented regarding laterality, location within the gray or white matter, and degree of gray versus white matter involvement. Because it was difficult to assess the severity of gray versus white matter lesions on CT scans, MR images were analyzed only for differential involvement of gray and white matter. The criteria for diagnosing gray matter involvement included small lesions located either entirely within the gray matter or a larger lesion predominantly in the cortex with the epicenter in the gray matter. The following characteristics were also noted: pattern of enhancement, evidence of hemorrhage, and late changes (atrophy and encephalomalacia). The interval between radiation therapy and the detection of imaging evidence of temporal lobe changes was also recorded. In patients with more than one imaging study, the morphology of the lesions was correlated with time course.

MR imaging was performed using either a 1.0-T scanner (Magnetom Expert; Siemens Medical Systems, Iselin, NJ) or a 1.5-T scanner (Vision; Siemens Medical Systems). Spin-echo technique was used in the axial and coronal planes. T1-weighted (TR/TE, 700/15) and T2-weighted (2730/80) MR images were obtained in the axial plane. T1-weighted MR images were also obtained in the coronal plane. After injection of gadopentetate dimeglumine (0.1 mmol/kg body weight; Magnevist; Schering, Berlin, Germany), axial and coronal sections were obtained. Slice thickness for axial and coronal images was 5 mm with a 2-mm interslice gap. T1-weighted MR images were obtained with two excitations, and T2-weighted images were obtained with one excitation. The imaging matrix was 192 × 256 in both scanning planes and sequences. CT was performed in the axial plane after the injection of 80 mL of iopromide (370 mg I/mL) on a 1200 scanner (Picker International, Cleveland, OH) or a HiSpeed scanner (General Electric Medical Systems, Milwaukee, WI). Scan thickness was 3 mm at the skull base and 5 mm above and below the skull base.

Results	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResults <<DiscussionReferencesCITING ARTICLES

The mean interval between radiation therapy and the onset of temporal lobe changes revealed on imaging was 5.4 years (range, 7-160 months).

On CT, 12 patients (35%) had unilateral lesions and 22 (65%) had bilateral lesions. The most common CT finding was lesions with ill-defined contrast enhancement associated with cerebral edema (Fig. 2A,2B,2C). This pattern was observed in 27 patients (79%). The next most frequently encountered CT appearance was solid enhancement, observed in six patients (18%) (Fig. 3A,3B). The ring-enhancement pattern was visible in one patient (3%) (Fig. 4). Five patients (15%) had lesions associated with a mid line shift.

View larger version (201K)

Fig. 2A. —51-year-old man with radiation-induced bilateral temporal lobe changes. Axial enhanced CT scan shows patchy enhancement and edema in right temporal lobe. Left temporal lobe is affected to lesser degree.

View larger version (212K)

Fig. 2B. —51-year-old man with radiation-induced bilateral temporal lobe changes. Axial enhanced T1-weighted spin-echo MR image shows extensive lesion in right temporal lobe (asterisk). Note preferential gray matter lesion in left temporal lobe (arrow) that may be seen in patients with nasopharyngeal carcinoma treated with radiation therapy.

View larger version (209K)

Fig. 2C. —51-year-old man with radiation-induced bilateral temporal lobe changes. Coronal enhanced T1-weighted spin-echo MR image shows gray matter lesion (white arrow) in left lobe. Note large lesion (asterisk) in right temporal lobe, white matter edema (star), and shift of midline (black arrow). Also note intracerebral lesions note associated with nasopharyngeal or extradural mass.

View larger version (174K)

Fig. 3A. —49-year-old man with radiation-induced unilateral temporal lobe changes. Axial enhanced CT scan obtained during assessment of nasopharynx shows ill-defined contrast enhancement in left temporal lobe.

View larger version (173K)

Fig. 3B. —49-year-old man with radiation-induced unilateral temporal lobe changes. Axial enhanced CT scan of brain obtained 1 hr after A shows solid pattern of enhancement (arrow).

View larger version (200K)

Fig. 4. —61-year-old woman with radiation-induced bilateral temporal lobe changes. Axial enhanced CT scan shows ring enhancement in left temporal lobe (asterisk). Note solid enhancement pattern on contralateral side (star).

Twenty-six patients underwent MR imaging (32 examinations). Eleven patients (42%) had unilateral abnormalities and 15 patients (58%) had bilateral temporal lobe changes. Simultaneous gray and white matter lesions were noted in 17 patients (65%), and nine patients (35%) had lesions localized to the gray matter. In temporal lobes with simultaneous gray and white matter lesions, more extensive involvement was noted in the white matter (Fig. 5A,5B). Of the subset of 22 temporal lobes with simultaneous gray and white matter changes, six (27%) showed more extensive gray matter involvement, and 16 lobes (73%) showed more severe white matter lesions. White matter lesions were characteristically associated with gross edema, but gray matter abnormalities were usually accompanied by less edema (Fig. 5A,5B). Furthermore, five temporal lobes showed enhanced gray matter lesions with no associated white matter edema.

View larger version (196K)

Fig. 5A. —59-year-old man with gray and white matter lesions, resulting in subsequent cerebral atrophy. Axial enhanced T1-weighted spin-echo MR image shows right (arrowheads) and left (star) temporal lobe lesions.

View larger version (210K)

Fig. 5B. —59-year-old man with gray and white matter lesions, resulting in subsequent cerebral atrophy. Axial enhanced T1-weighted spin-echo MR image obtained 10 months after steroid therapy shows decrease in lesion size. Note subarachnoid space (asterisks) and onset of cerebral atrophy with temporal horn dilatation (stars). Gray matter enhancement may persist for years.

MR imaging showed variable temporal lobe abnormalities after radiation. These lesions typically had irregular margins. Large lesions may show central areas with no enhancement, presumably caused by necrotic debris (Fig. 5A,5B). Large lesions may also show significant space-occupying effect, resulting from white matter edema. Seven patients (27%) had MR imaging evidence of subacute or chronic hematoma (Fig. 6A,6B). Five of these seven patients showed low-signal-intensity changes on T1- and T2-weighted MR images. Two patients had high signal intensity on T1- and T2-weighted MR images (Fig. 7A,7B,7C). In both of these patients, low signal intensity on T1- and T2-weighted MR images was present simultaneously. In the 47 patients examined, three (6%) had cerebral atrophy and one each (2%) had microcystic features (Fig. 7A,7B,7C) and macrocystic encephalomalacia on follow-up studies.

View larger version (215K)

Fig. 6A. —62-year-old man with chronic hematoma associated with radiation-induced temporal lobe changes. Coronal unenhanced T1-weighted spin-echo MR image shows left temporal lobe swelling (star) and low-signal-intensity curvilinear hemosiderin (arrows).

View larger version (188K)

Fig. 6B. —62-year-old man with chronic hematoma associated with radiation-induced temporal lobe changes. Coronal T2-weighted spin-echo MR image obtained 2 weeks after A shows better delineation of chronic hematoma (arrow) and persistent edema (E). Hemosiderin is often seen on T2-weighted spin-echo MR images in patients with moderate to extensive lesions.

View larger version (215K)

Fig. 7A. —49-year-old woman with bilateral radiation-induced temporal lobe changes showing subacute hemorrhage and encephalomalacia. Axial enhanced T1-weighted spin-echo MR image shows high-signal-intensity foci (arrows) in right temporal lobe. Note intermediate signals in left temporal lobe (star).

View larger version (209K)

Fig. 7B. —49-year-old woman with bilateral radiation-induced temporal lobe changes showing subacute hemorrhage and encephalomalacia. Axial T2-weighted spin-echo MR image shows low and high signals on right indicating blood products. Note cystic changes (star) in left temporal lobe and associated edema.

View larger version (206K)

Fig. 7C. —49-year-old woman with bilateral radiation-induced temporal lobe changes showing subacute hemorrhage and encephalomalacia. Coronal enhanced T1-weighted spin-echo MR image shows cystic encephalomalacia (white star) in left temporal lobe. Note inflammatory changes (black star) in sphenoid sinus and slough (arrow) in nasopharynx.

Of the 47 patients in this series, 28 (60%) had bilateral temporal lobe lesions. In this subset of 28 patients, 24 (86%) had asymmetric involvement. Nineteen patients (40%) had unilateral temporal lobe changes, and of these, 11 (59%) had nasopharyngeal tumor corresponding to the side of temporal lobe involvement.

Thirteen patients underwent MR imaging and CT. Four of these patients had CT and MR imaging studies performed within a 3-month period and these examinations were compared. In all four patients, the typically ill-defined lesions seen on CT were better delineated with contrast-enhanced MR imaging. Two patients had gray matter lesions on MR imaging, but these lesions were not revealed on CT (Fig. 2A,2B,2C).

Thirty-one patients had two or more studies. In this subset of patients, 17 (55%) showed stable findings. The average interval between examinations in patients with stable findings was 2 years 7 months (range, 10-56 months). The following characteristics were noted in the remaining 14 patients (45%): five had progression to atrophy and cystic encephalomalacia, three had rapid clearing of edema after corticosteroid therapy, three had enlarging gray matter lesions that involved white matter, two with small gray matter lesions had spontaneous resolution, one had an enlarging hematoma, and one with an initial unilateral lesion progressed to bilateral involvement.

Discussion	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResultsDiscussion <<ReferencesCITING ARTICLES

Radiation therapy techniques vary from institution to institution. Our patients were treated using a modification of the technique of Ho [11] and Tsao [12]. The treatment field includes the sphenoid and posterior ethmoid sinuses; the posterior half of nasal fossa and maxillary sinus; the orbit posterior to the globe; the foramen rotundum, ovale, and jugular foramen; the anterior half of the petrous temporal bone; the basiocciputal and petrooccipital fissure; the palatal and pterygoid muscles; and the oropharynx. The standard radiation dose for the treatment of nasopharyngeal carcinoma is 66-70 Gy, and for recurrence, another dose of up to 50 Gy may be given. In the previously mentioned technique, the 90% isodose curve includes a portion of the inferomedial aspect of both temporal lobes.

Ionizing radiation damages deoxyribonucleic acid and the injured cells perish only after failed attempts at mitosis. Neurons are radioresistant because they do not divide but they die as a consequence of injury to the oligodendrocytes and endothelial cells. The rate at which injury develops is related to the tissue proliferative activity. Tissues in the central nervous system, with slow cellular turnover, typically show delayed changes. In our patients, the mean time between radiation therapy and the first sign of temporal lobe changes on imaging was 5.4 years. It can be assumed that temporal lobe changes predate imaging diagnosis for an unknown period. Therefore, the actual onset of temporal lobe changes can be much earlier. Our data on the onset of temporal lobe changes are in keeping with the findings of earlier publications [7, 10].

For the treatment of nasopharyngeal carcinoma, the radiation dose to both sides of the nasopharynx is approximately equal. In patients with large nasopharyngeal lesions, a parapharyngeal boost (typically 10 Gy) is given to the involved side. However, the radiation dose to both temporal lobes remains approximately equal. It is unknown why differential involvement of the temporal lobes occurs in a high percentage of patients. In this study, 35% of patients who underwent CT and 42% of patients who underwent MR imaging had unilateral abnormalities. These figures are similar to those of a previous report in which half the patients with temporal lobe changes presented with unilateral abnormalities [10]. The most common CT appearance of temporal lobe changes was cerebral edema with patchy contrast enhancement. Delayed scans may show better demarcation of radiation injury (Fig. 3A,3B). Lesions showing a solid pattern of enhancement are less common. The ring-enhancement pattern, although well documented in the literature, was rare in our study [13,14,15].

The differential involvement of gray and white matter is best revealed on MR imaging. In this study, simultaneous gray and white matter lesions were revealed in 17 patients (65%), and nine patients (35%) had only small gray matter lesions. This high frequency of gray matter involvement in our study differs from that of studies that report radiation necrosis preferentially affects the white matter in patients with brain tumors [16, 17]. This observation may be explained by the fact that treatment for nasopharyngeal carcinoma inevitably includes within the radiation portal the medial and inferior portions of the temporal lobes (mainly gray matter).

Radiation therapy may induce mineralizing microangiopathy and telangiectasia in the brain [13]. Mineralizing microangiopathy appears as multiple punctate calcifications in the cerebral cortex, brainstem, or basal ganglia. This entity was not seen in our patients. Radiation therapy can also cause dilatation of capillaries in the white matter. Telangiectasia is thought to be caused by the development of collateral vessels and may result in subclinical hemorrhage [18]. In our study, seven patients (27%) who underwent MR imaging had chronic hematoma (Figs. 6A,6B and 7A,7B,7C). These patients showed low-signal-intensity lesions on T1- and T2-weighted MR images. Additionally, two patients also had concomitant high-signal-intensity foci on T1- and T2-weighted MR images.

It is difficult to describe the evolution of temporal lobe changes in this retrospective, descriptive study. The onset of temporal lobe changes cannot be accurately established because this complication is frequently asymptomatic. It can be assumed that pathologic changes preceded radiologic detection for an unknown period. Therefore, unfortunately, this study cannot accurately correlate the time interval (months or years) between radiation therapy and the appearance of specific imaging characteristics.

Two early signs of radiation injury exist: small gray matter enhancing foci and cerebral edema. These gray matter lesions may be seen 2 years after radiation therapy and may remit spontaneously after 1 year. The more commonly seen early lesion was edema. It should be noted that although cerebral edema is an early sign, it might persist for years after initial detection. Similarly, large gray matter lesions show persistent enhancement for 3-4 years. In general, radiation-induced changes in the temporal lobes are characterized by an indolent course. Eventually, some patients show cerebral atrophy 10 months after corticosteroid therapy [19] (Fig. 5A,5B). Cystic encephalomalacia may appear as a single large cyst or a collection of small cysts (Fig. 7A,7B,7C).

The differential diagnosis of temporal lobe changes includes intracranial tumor recurrence and hematogenous metastasis [20]. Most recurrent nasopharyngeal carcinomas with intracranial extension are entirely located in the extradural space. These recurrent tumors are rarely associated with cerebral edema. Therefore, an extraaxial mass and cerebral edema are suggestive of intracerebral invasion (Fig. 1). However, radiation-induced temporal lobe changes are intraaxial abnormalities, and cerebral edema is a common early sign. Radiation-induced changes are usually indolent and the radiologic signs often persist for years. Such an indolent course is unexpected in patients with recurrent tumor. Therefore, in the absence of a nasopharyngeal mass or epidural lesion, lesions in the inferomedial aspect of the temporal lobes are strongly suggestive of radiation-induced changes. The location of temporal lobe changes and tumor recurrence in relation to the skull base makes coronal enhanced MR imaging the technique of choice.

In general, intracerebral lesions in patients with a history of malignancy should be considered metastatic disease until proven otherwise. Radiation-induced necrosis shares many common features with metastatic lesions. However, hematogenous metastases in nasopharyngeal carcinoma are exceedingly rare [21]. In practice, cerebral metastasis is seldomly placed high on the list of differential diagnosis.

Positron emission tomography (PET) with ¹⁸F-fluorodeoxyglucose (FDG) may be used to identify radiation injury to the brain. PET shows considerably lower levels of FDG in areas of radiation-induced necrosis compared with areas of recurrent tumor. ²⁰¹Tl-chloride single-photon emission computed tomography (SPECT) can also be used to separate these entities. However, no significant difference exists in the abilities of FDG PET (sensitivity, 81%; specificity, 40%) and ²⁰¹Tl-chloride SPECT (sensitivity, 69%; specificity, 40%) to accurately separate tumor from radiation-induced necrosis [22]. Radiation injury may also be identified on MR spectroscopy. The earliest manifestation of radiation injury is a decrease in N-acetylaspartate, which may be evident before changes in choline or creatine [23,24,25]. Recurrent nasopharyngeal tumor with intracranial extension is not expected to show N-acetylaspartate levels because the tumor is not of neural origin. Therefore, the detection of N-acetylaspartate in a lesion indicates temporal lobe necrosis rather than intracranial nasopharyngeal tumor progression or recurrence, and MR spectroscopy has the potential to show the differences between radiation-induced changes and tumor progression [25].

The temporal lobe changes revealed in this study most likely represent radiation-induced necrosis. One of the weaknesses of this study is a lack of biopsy confirmation. The diagnosis of temporal lobe necrosis is based on the correlation of clinical and radiologic findings. The disparity between clinical and radiologic findings is noteworthy and highly suggestive of temporal lobe necrosis [26, 27]. Patients with gross radiologic abnormalities may have relatively mild symptoms. At presentation, 37 patients (79%) in this study were asymptomatic, whereas the remaining 10 had vague symptoms, such as forgetfulness, headache, and mild psychomotor retardation. In most cases, pathologic proof of radiation-induced necrosis is not required [28]. Our patients were closely followed up clinically, and correlation with radiologic findings did not justify open brain biopsy. This strategy can be adopted when no mass is noted in the skull base, radiologically or endoscopically. When necessary, a biopsy can be performed, as was the case in one patient.

The treatment of temporal lobe changes depends on clinical assessment. In asymptomatic patients with mild radiologic changes, no corticosteroid therapy is required because of potential, serious hazards. Corticosteroid treatment is instituted in symptomatic patients and those with more pronounced radiologic findings. In more than one third of these patients, objective clinical and radiologic response may be obtained [27]. Surgical removal of necrotic lesions is the definitive way of achieving long-term control in patients with debilitating symptoms; however, this method may not be suitable for patients with bilateral temporal involvement [29].

In summary, radiation therapy for nasopharyngeal carcinoma invariably includes, within the treatment field, the inferomedial portions of the temporal lobes. Although the radiation dose is approximately equal on both sides, many patients have unilateral lesions. Temporal lobe changes can involve the gray and white matter simultaneously or the gray matter alone; however, isolated white matter lesions are rare. White matter lesions are typically associated with edema, and gray matter lesions have less or no associated white matter edema. Late temporal lobe changes include cerebral atrophy and microcystic or macrocystic encephalomalacia. PET scans can provide additional information; however, if this technique is unavailable, then these findings can be sufficient to separate radiation-induced changes from intracranial tumor recurrence.