The effect of fractionated doses of radiation on mouse spinal cord

To determine: (a) the dose-response relationship and latent time to paralysis following fractionated doses of radiation in mice, (b) the values of parameters for isoeffect curves, and (c) whether these parameters depend on the size of dose per fraction and the severity of injury.

The spinal cords (T9-L5) of 608 C3Hf/Sed/Kam mice were irradiated with fractionated doses of x-radiation. Three levels of neurological damage were used to grade the spinal cord response. Animals which did not develop paralysis were observed for at least 18 months after irradiation. The fractionated schedules consisted of either 2, 3, 4, 6, 10, or 20 fractions in addition to single doses. For the fractionated regimes the daily fraction size ranged from 2 Gy to 24 Gy, and for single doses the range was 12 Gy to 52 Gy. Both the latent time to paralysis and the incidence of paralysis were considered as endpoints. For analysis of the sparing associated with fractionation, the dose points were divided into two groups: a "low damage" group consisting of doses of near or less than the ED50 at 450 days and a "high damage" group consisting of doses much larger than the ED50 at 450 days in which there was 100% incidence of paralysis.

The latent time depended on the radiation dose; for each fixed fraction number the latent period became progressively shorter with higher total doses. Differences in histology in fractionation sensitivity are observed between the two groups. The low damage data in each fractionation treatment are the important data in the analysis of long-term incidence of paralysis. On the other hand, the high damage data were emphasized for the analysis of latency. Three statistical methods (mixture model, Cox model, and Fe-plot) were used to fit the linear-quadratic dose response model and the "Nominal Standard Dose" (NSD) model. The values of the parameters of these two models depended on the effect evaluated; the latent interval from the high damage region being not very fractionation-dependent, whereas, the incidence of paralysis from the low damage fractionation regimens was strongly dependent on dose per fraction. Specifically, the alpha/beta ratios for latency were large (e.g., 17 to 57 Gy) when fractionation schemes in the high damage region were emphasized. If data from the fractionation schemes in the lower damage region with fraction size less than 15 Gy were emphasized, the alpha/beta ratios for incidence of paralysis were 3.3 (1.8, 6.0, 95% C.I.), 4.1 (2.8, 5.5), and 4.4 Gy derived by the mixture, Cox, and Fe-plot models, respectively. These "low damage" alpha/beta ratios were similar for all levels of injury from mild to complete paralysis, and are those which are more relevant to clinical radiotherapy. The coefficients for the "Nominal Standard Dose" formula in the present study were 0.33 +/- 0.01 (s.e.) (by the Strandqvist-type plot), 0.38 (the Cox model), or 0.40 (the mixture model) for level 2 injury at 450 days.

The values of parameters in the isoeffect models were different when the data analyzed were derived from regimens using fractionated low or high damage doses.