Multifocal eosinophilic granuloma and the cerebellum

Abstract

AdQmatq et al [I] reported seeing hypodense areaz-Ln the cerebellum on computed t o m o e h y (CT) of the brain in two patients with multifocal eosino~hilic granuloma. We have cared for two patients with similar lesions whose findings help answer some of the questions raised by Adornato and his colleagues.A boy developed proptosis at 2 years of age. Biopsy of an orbital mass revealed eosinophilic granuloma, and chemotherapy with vincristine and prednisone was begun. Six months later he developed gait ataxia. A C T scan showed symmetrical hypodense areas in the cerebellum in the region of the dentate nuclei. Hounsfield numbers in the lesions were between 9 and 31. The orbital mass persisted during this period, and radiation therapy with 1,000 rads was administered to the left frontal and orbital areas. C T scans repeated over the next two years showed no changes in the cerebellar lesions.A ll/z-year-old boy was found by biopsy to have an eosinophilic granuloma of the left orbit. He developed multiple skull lesions and received 4,800 rads of radiation therapy and chemotherapy over the next three years; the radiation ports were restricted to the orbits and temporal areas. A progressive truncal ataxia occurred, and a C T scan showed bilateral hypodense areas in the cerebellum in the region of the dentate nuclei. A course of radiation therapy was administered to the posterior fossa. Over the next three years, CT scans showed no progression of the hypodense areas. The ataxia has steadily progressed.Cerebellar disturbance in multifocal eosinophilic granuloma has been associated with dentate calcification and atrophy of the cerebellum [ 11, xanthomatous granulomas [ 5 ] , and demyelinating lesions 12, 31 with or without concomitant xanthomatous infiltration. Adornato et al [ 13 suggested that the hypodense cerebellar lesions in their patients were unrelated to prior radiation therapy. The presence of the cerebellar lesions in our first patient preceded irradiation, and the radiation therapy administered to our second patient did not include the posterior fossa until after the cerebellar lesions were demonstrated. Solely granulomatous infiltration also seems an unlikely explanation for the lesions, as histologically proved areas of granulomatous tissue have been shown to be isodense or hyperdense on C T scan [ 5 ] . The orbital lesions in our patients were of increased density. Adornato and his associates concluded that the areas of hypodensity in their patients were regions of "demyelination coexistent with histiocytic infiltration" [ 1 I. We also favor demyelination but think it may occur without coexistent infiltration of histiocytes [2, 31. Subsequent to the discovery of cerebellar lesions in our patients there has been no recurrence of multifocal eosinophilic granuloma; the cerebellar symptoms have persisted or progressed, but the lesions have remained unchanged on CT scan.The wisdom of treating such lesions with radiation therapy must also be questioned. Chemotherapy and irradiation in leukemic children have produc...