PURPOSE/OBJECTIVE: Some malignant brain tumors shed cells into the cerebrospinal fluid (CSF). These tumors may implant throughout the neuroaxis via the CSF. With the placement of a ventriculoperitoneal (VP) or ventriculoatrial (VA) shunt, tumor cells free-floating in the CSF may be carried through the shunt to the remainder of the body. Mechanical filtration devices to prevent this are not reliable. We report the development of a new device capable of reducing the incidence of shunt promoted metastasis. MATERIALS & METHODS: The device exposes the draining CSF, as it passes through a baffle system, to a localized high-intensity radiation field adequately shielded from surrounding normal tissue. The prototype consists of geometrically fixed iodine-125 (125I) sources. The device accommodates the maximum CSF flow rate of 500 ml/24 hours. Radiation exposure to clonogenic cells occurs as they transit through the baffle system. Since the volume of the prototype device is 14 ml, a tumor cell floating through the device will be exposed to radiation for 40 minutes. Utilizing the human medulloblastoma cell line D425 MED, a limiting dilution clonogenic assay was performed. Suspensions of tumor cells in liquid medium were pumped through the device at the maximum anticipated CSF production rate of 0.35 ml/min. After the cells, with their tissue culture medium, were received from the device, a series of nine 5-fold dilutions were prepared from the suspensions which initially contained 10(6) tumor cell/ml. Plates were then incubated and growth was demonstrated by visual scoring of colonies of more than 20 cells. Limiting dilution data analysis was performed. Radiation surveys of the fully loaded (approximately 1.8 Ci) 125I prototype were conducted. A well calibrator was used to measure the activity of the fully loaded device. RESULTS: When the device was loaded with 125I seeds providing a dose of 364-479 cGy the probability of clonogen survival was 0.033. Radiation exposure levels at the exterior surface of the shielded device were in the range of 2-5 mR/hr and thus fell within guidelines for acceptable normal tissue exposure. Attenuation of radiation by the shielding case for the fully loaded device was 10(-5). CONCLUSION: The device kills medulloblastoma cells as they are pumped through it. If the risk of metastasis is linearly related to the number of clonogenic cells, then the device would, we infer, reduce the risk of shunt-born metastasis by a factor of 0.033 and merits further investigation.