Radiation necrosis in the temporal pole and lateral temporal cortex, presenting with a deficit of semantic memory
When this 57 year old, high-functioning gentleman initially presented to the Neurology clinic, the impression was that of an early degenerative process affecting the medial and lateral temporal lobes, for which an outpatient evaluation was initiated. He received an elective MRI study two months after his initial presentation, and was called back into the hospital later that day for urgent admission to evaluate the finding of a very large, inhomogeneously enhancing mass in the left temporal lobe, with surrounding edema, causing early transfalcine herniation. He was treated empirically with dexamethasone for the extensive edema, and carbamazepine, as on review of his history the spells of being unable to speak and "blanking out" followed by confusion were suspicious for complex partial seizures. The working diagnosis at that point was a metastatic lesion from his previous T4 squamous cell carcinoma which had extended into the middle ear and eroded the left temporal bone at the time of diagnosis three years prior. Although it was known that his temporal lobe had been in the radiation field, the extremely malignant appearance and size of the lesion precluded an assumption of radiation necrosis on clinical and radiologic grounds.
The Neurosurgery and Otorhinolaryngology services were consulted. There was no local recurrence seen in the nasopharynx, and the temporal bones appeared intact on CT with fine cuts made through the temporal region, suggesting that local extension into the temporal lobe was unlikely. CT of the chest and abdomen revealed a pleural-based mass in the left lower lobe of the lung which was suspicious for neoplasm. On discussion with the Pulmonary service, this was thought to represent a second primary neoplasm, not a metastasis. We decided to proceed with primary surgical resection of the patient's brain mass, as the size of the lesion and the amount of mass effect it was producing left no viable therapeutic alternatives. Our differential diagnosis before the patient went to surgery included a metastatic lesion from either his old or his new primary, a primary brain tumor arising in the radiated field, or radiation necrosis. We did not pursue functional or metabolic imaging studies such as SPECT, PET or MR spectroscopy prior to surgery, as we did not feel that a negative study would have precluded the need for surgery to palliate the effects of the patient's brain mass.
The mass was resected in toto two weeks later (delayed at the patient's request). Histopathological analysis showed changes consistent with radiation necrosis, and no evidence of tumor cells. The patient did very well clinically; his deficits remained stable to mildly improved postoperatively, and he returned to work full time. He pursued evaluation of the pleural-based lung nodule on an outpatient basis, and a fine needle aspiration of the mass was inconclusive. On follow-up in the Neurology Clinic three months later, he reported doing well, with stable neuropsychological deficits. He had no further spells on empiric carbamazepine. One month later he presented to the emergency room with abdominal distension and ileus. Exploratory laparotomy showed extensive metastatic disease of the liver, peritoneum and omentum. The patient elected palliative care, and passed away at his home a few weeks later. No autopsy was performed.
The patient's presenting symptom was an inability to name personal acquaintances and public figures, much more so than his mild difficulties in general word retrieval, which he himself had not noted. Also, he complained of difficulty with memory for recent events. Neuropsychological testing corroborated these two complaints. On word list learning, he exhibited a minimal learning slope and was impaired on both free and cued recall; recall for a prose narrative and for visual-spatial stimuli was severely impaired. This pattern is typically seen in dysfunction of the hippocampal memory system (for a good review, see Squire, 1992). Although the patient's medial temporal lobe was relatively spared, it must be assumed that there was a significant amount of hippocampal disconnection due to the lesion itself and the massive edema it caused. Most of the tissue destruction, though, was located in the temporal pole and lateral temporal cortex. While the exact function of this cortical region remains under investigation, it has been implicated to play a role in semantic memory. The left temporal pole, more specifically, has been found to be involved in the naming of unique entities. Our patient showed deficits on testing that were quite consistent with the types of difficulties described with lesions or processes involving this region.
Degradation of semantic knowledge in temporal lobe disorders has been studied most extensively in the syndrome of semantic dementia, characterized by a progressive loss of meaningful speech and a gradual deterioration of knowledge about the world (e.g., Hodges et al., 2000, Murre et al., 2001). In this disorder, there is early difficulty with naming, but patients also lose information about other aspects of things (for example the use, color, and detailed appearance of objects). Additionally, they lose information of irregular grammatic forms, pronunciation and spelling, and tend to "regularize" when reading or speaking (for example, "the boy falled" instead of "the boy fell"), whereas their general use of grammar and syntax remains intact. Typically, they lose the ability to recognize the core concept of things, exhibit intrusion of information about semantically related objects, and later regress toward category exemplars (for example, the difference between a swan and a pelican becomes very fuzzy, as both are large water birds, and eventually both are thought of as birds). Another important aspect of semantic dementia is the early loss of semantic autobiographical content as well as of highly consolidated memories, especially childhood memories, as these are believed to become part of the semantic store once they become independent of the hippocampus.
Patients with semantic dementia typically have atrophy of the lateral temporal cortex and the temporal pole, with relative sparing of the medial temporal lobe (Garrard et al., 1997). In a recent morphometric MRI study, the most consistent locus of atrophy was the left polar and inferior temporal lobe (Mummery et al., 2000). Degradation of semantic knowledge and difficulty with naming, very much like the deficits described for semantic dementia, have also been observed in patients with other temporal lobe lesions. These deficits have been described in patients with bilateral temporal lesions that involve the lateral temporal cortex and temporal pole, due to herpes simplex encephalitis or epilepsy surgery (Schmolck et al., 2002). Multiple functional imaging studies with PET and MRI have shown activation in the lateral temporal cortex and fusiform gyrus during tasks requiring semantic judgements or other lexico-semantic processing (e.g., Vandernberghe et al., 1996; Demonet et al., 1992; Cabeza & Nyberg, 2000). Regarding the specific role of the left temporal pole in semantic memory, it appears that regions involved in naming (and knowledge) for general entities are located more posteriorly, while retrieving the names for unique persons (e.g., Marilyn Monroe) or for names of unique landmarks (e.g., the Louvre) appear to be subserved by regions located more anteriorly in the left temporal pole (Grabowski et al., 2001).
Our patient's difficulties were in accord to the proposed functions of the anterior lateral temporal cortex areas. The most frustrating problem for him was the failure to retrieve names for unique entities. This impairment was demonstrated during formal testing when he was given the Famous Persons Test. He had some further impairments on testing that are seen with lateral cortex lesions and the degradation of semantic knowledge: mild impairments in naming, especially for lower frequency items; normal phonemic (FAS) fluency but impaired category fluency (which requires intact semantic processing; in normal individuals, phonemic fluency is more difficult than category fluency); and relatively low scores on the verbal knowledge subscales of the IQ testing (in relation to his other subscales).
Delayed radiation necrosis can be seen early (more than 10 weeks postradiation) or late (several months to years after radiation). The histological changes in early delayed radiation injury consist of demyelinating plaques and occasionally intense microglial and astrocytic reactions, with perivascular lymphocytic and plasma cell infiltration. Late delayed injury can include variable glial reactions, perivascular hyaline deposits with frequently severe vascular calcification, occasionally intense acellular fibrosis, and infiltrates of immune cells.
While in our patient's case, the differential diagnosis of tumor recurrence versus radiation necrosis did not play a major role in management due to the malignant appearance on imaging and the mass effect of the lesion, it is an important distinction in many other cases and thus merits more detailed discussion. In our patient, the intense radiation protocol with the "Peacock" paradigm three years prior to presentation subjected the left temporal lobe to significant radiation, which we believe eventually led to the massive tumor-like necrosis seen on MR imaging. More commonly, the differential diagnosis of tumor versus radiation necrosis is entertained in a patient with a previously irradiated brain tumor, most commonly a glioma, and new MRI abnormalities in the tumor bed on follow up imaging. Classically, FDG-PET and SPECT have been used to differentiate between radiation necrosis and tumor progression, but multiple reports in the literature now describe FDG hypermetabolism on PET in histologically proven radiation necrosis (e.g., Fischman et al., 1997; Schlemmer et al., 2002). Even the usefulness of stereotactic biopsy for diagnosis has been questioned, as there is always the possibility of tumor recurrence in conjunction with radiation necrosis in the same field (Ehrenfeld et al., 2002).
Recently, magnetic resonance spectroscopy (MRS) has been suggested to be useful in making make the important distinction between tumor and radiation necrosis (Schlemmer et al., 2002; Rock et al., 2002). Several metabolic ratios are described to be differentially affected by either tumor cells or tissue necrosis. In general, moderately increased choline peaks and reduced N-acetylaspartate (NAA) signal intensities indicate low grade gliomas, high grade tumors are characterized by distinctly higher choline peaks and even lower NAA signals, and the presence of lipid signals indicates tissue necrosis. In a recent article (Rock et al., 2002), the authors were able to correctly predict the histopathology of the subsequently resected specimen in the case of pure radiation necrosis or pure tumor recurrence using several metabolic ratios. However, the spectral patterns were less definitive when tissues composed of varying degrees of mixed tumor and necrosis were examined. The animated discussion in the journal (Neurosurgery) following this article illustrates the extent of the problem and the wide interest clinicians take in it. The consensus among responders' comments appeared to be that most neurosurgeons and neuroradiologists value the additional information that MRS supplies, but currently would favor resection for definitive diagnosis in most cases, especially in light of the indeterminate findings in mixed lesions. The hope is that refinement of the technique might be of greater impact in the future.
We thank the staff of the Houston Veterans Affairs Medical Center for their assistance with this patient's care. Most of all, we thank the patient, who gave consent for the posting and discussion of his case.
-- Dennis R. Mosier, M.D., Ph.D.
Email comments: