Volume effects in the irradiated canine spinal cord: do they exist when the probability of injury is low?

PURPOSE: The purpose of this study was to investigate volume effects in the irradiated canine spinal cord. MATERIALS AND METHODS: Eighty-nine beagle dogs were given 44-84 Gy photons in 4 Gy fractions to 4 or 20 cm lengths of thoracic spinal cord. As controls, 36 dogs were given 60-84 Gy in 2 Gy fractions to a 20 cm length of spinal cord and six dogs were unirradiated. Dogs were evaluated for clinical signs, and after euthanasia, for occurrence of gross lesions, severe lesions of massive hemorrhage, white matter necrosis and/or parenchymal atrophy and mild lesions of focal fiber loss. White matter vacuoles, meningeal thickness and dorsal root ganglia lesions were quantified. Data were analyzed to test for an effect of volume on dose-response curves. RESULTS: Significant volume effects were found between 4 and 20 cm lengths of irradiated spinal cord for gross lesions, severe lesions and mild lesions (8.3-15.0 Gy difference at the ED50 level). The ED50 in 4 Gy fractions for severe lesions was 56.9 Gy (95% CI 53.1-60.6) for 20 cm and 68.8 Gy (95% CI 64.5-75.1) for 4 cm fields. Significant improvements in the fit of data to dose-response curves resulted when using models with either parallel or non-parallel curves, but in either case an appreciable difference existed between curves at low probabilities of injury. Volume effects were present for meningeal thickness and slopes of dose-response curves were different. Clinical signs correlated well with severe lesions for 20 cm (ED50 = 54.0 Gy), but not for 4 cm fields (ED50 = 77.6 Gy). CONCLUSIONS: Volume effects exist for the occurrence of pathologic lesions in irradiated canine spinal cord. Clinical compensation for pathologic lesions occur at small, but not large irradiated volumes. There is insufficient data to support a decreased slope of dose-response curves with decreased volume. Volume effects estimated at the 50% level of spinal cord injury could also hold at low probabilities of injury characteristic of the clinic.