Probabilities of radiation myelopathy specific to stereotactic body radiation therapy to guide safe practice

Purpose: Dose-volume histogram (DVH) results for 9 cases of post spine stereotactic body radiation therapy (SBRT) radiation myelopathy (RM) are reported and compared with a cohort of 66 spine SBRT patients without RM. Methods and Materials: DVH data were centrally analyzed according to the thecal sac point maximum (Pmax) volume, 0.1- to 1-cc volumes in increments of 0.1 cc, and to the 2 cc volume. 2-Gy biologically equivalent doses (nBED) were calculated using an α/β = 2 Gy (units = Gy_2/2). For the 2 cohorts, the nBED means and distributions were compared using the t test and Mann-Whitney test, respectively. Significance (P<.05) was defined as concordance of both tests at each specified volume. A logistic regression model was developed to estimate the probability of RM using the dose distribution for a given volume. Results: Significant differences in both the means and distributions at the Pmax and up to the 0.8-cc volume were observed. Concordant significance was greatest for the Pmax volume. At the Pmax volume the fit of the logistic regression model, summarized by the area under the curve, was 0.87. A risk of RM of 5% or less was observed when limiting the thecal sac Pmax volume doses to 12.4 Gy in a single fraction, 17.0 Gy in 2 fractions, 20.3 Gy in 3 fractions, 23.0 Gy in 4 fractions, and 25.3 Gy in 5 fractions. Conclusion: We report the first logistic regression model yielding estimates for the probability of human RM specific to SBRT.