Spine radiosurgery causes a rapid dose fall-off within the spinal cord. The tolerance of partial volume of the spinal cord may determine the extent of clinical application. The study analyzed the partial volume tolerance of the human spinal cord to single fraction radiosurgery. A total of 230 lesions with spine metastases in 177 patients were treated with radiosurgery with single fraction of 8 to 18 Gy, prescribed to the 90% isodose line that encompassed the target volume. Spinal cord volume was defined as 6 mm above and below the radiosurgery target volume. Spinal cord dose was calculated from the radiation dose/spinal cord volume histogram and correlated with clinical/neurological status and radiographic studies. Median follow-up was 6.4 months (range, 0.5-49 months). The 1-year survival rate was 49%.

The average spinal cord volume defined at the treated spinal segment was 5.9 +/- 2.2 mL. The average dose to the 10% spinal cord volume was 9.8 +/- 1.5 Gy, calculated from the dose-volume histogram in the group of 18 Gy prescribed dose. The spinal cord volume that received higher than 80% of the prescribed dose was 0.07 +/- 0.10 mL, which represented 1.3 +/- 1.8% of the cord volume. Among the 86 patients who survived longer than 1 year there was 1 case of radiation-induced cord injury after 13 months of radiosurgery. There were no other cases of spinal cord sequelae.
CONCLUSIONS. Whereas the maximum spinal cord tolerance to single-dose radiation is not known, partial volume tolerance of the human spinal cord is at least 10 Gy to 10% of the spinal cord volume defined as 6 mm above and below the radiosurgery target.

The procedure of spine radiosurgery has been previously described. The physical and dosimetric characteristics of shaped beam radiosurgery Novalis system (BrainLab, Germany) have been reported. This radiosurgery system uses frameless image-guided positioning and targeting. Immobilization was achieved primarily by using the Bodyfix device (Medical Intelligence, Schwabmunchen, Germany) with vacuum bags. Infrared reflective markers were placed on the skin. CT simulation was performed with intravenous contrast in 2-3 mm slices without spacing. By using the dedicated planning system with BrainScan planning computer (BrainLab), image fusion was routinely performed with simulation CT and MR images. The radiosurgery target volume and spinal cord were delineated. Radiosurgery used multiple (usually 7-9 beams) coplanar intensity-modulated radiation beams to minimize the dose to the critical organs. For treatment, image-guided repositioning was achieved by using an infrared marker and image fusion of internal bony structures. Before the delivery of radiation, orthogonal portal films were obtained for final verification of the isocenter.

All patients received single-dose radiosurgery to the involved spine only. The target volume included the involved vertebral body and pedicles. When there was a paraspinal or epidural component the involved spine and the gross visible tumor were included in the target volume. Ultimately, the radiosurgery planning target volume (PTV) was the same as the clinical target volume (CTV) including the gross tumor and the involved spine. There was no margin for planning of radiosurgery. The method of target delineation is diagrammatically illustrated with examples of the actual isodose distribution in Figure . There can be several different scenarios of spine involvement. Vertebral body involvement was the most common type of spine metastasis (Fig. A). In this case, the entire involved vertebral body and both pedicles were treated with radiosurgery. When the metastasis involves the dorsal elements (spinous process and laminae), the target included only the dorsal elements as in Figure C. Involvement as in Figure B was treated either to the entire bony element (vertebral body and dorsal elements) or in the same fashion as Figure A,C but with generous clinical margin (dotted line). No additional margin was given to the planning tumor volume. Spinal cord volume was consistently defined as the volume extending from 6 mm above to 6 mm below the radiosurgery target. The target tumor and the spinal cord were delineated by fusion of contrast-enhanced simulation CT images with T1-weighted MR images with and without gadolinium contrast and T2-weighted MR images. In the lumbosacral region below the cauda equina, the volume of filum terminale was defined in the same method as the spinal cord. Radiosurgery doses ranged from 8 to 18 Gy in a single fraction. The radiation dose has been consistently prescribed to the 90% isodose line that encompassed the periphery of the target tumor to have the same estimate of the spinal cord dose. The spinal cord constraint has been 10 Gy to the 10% partial volume of the spinal cord, based on the observation of a Phase I study. The main criterion of radiosurgery dose selection was the spinal cord dose constraint. When the spinal cord criterion was not met, the tumor coverage was compromised or the radiosurgery dose was reduced. In cases where the target coverage was important, the criterion was relaxed based on the patient's neurologic or general condition. The radiosurgery doses were 10 Gy in 37 lesions, 12-13 Gy in 44 lesions, 14 Gy in 48 lesions, 16 Gy in 62 lesions, and 18 Gy in 39 lesions.