Dose Impact in Lung Fibrosis Following Lung SBRT: Statistical Analysis and Geometric Interpretation
V Yu*, A Kishan, P Lee, D Low, D Ruan, UCLA School of Medicine, Los Angeles, CASU-C-141-1 Sunday 1:00PM - 1:55PM Room: 141
Purpose: To test the hypothesis that certain dosimetric parameters prognosticate treatment-induced lung fibrosis. To obtain estimates for most prognostic dose values and to utilize them in developing new planning strategies to reduce injury in future SBRT.
Methods: Follow-up CT scans at 6 and 12 months were acquired from patients treated with SBRT (18Gy/Fx*3Fx or 12.5Gy/Fx*4Fx) for stage-1 primary lung cancers or oligometastic lesions. Fibrosis regions were identified in these scans and propagated to the planning CT coordinates by rigidly registering the follow-up and the planning CTs. Among a cohort of 8 properly registered cases, a bimodal behavior was repeatedly observed from the probability distribution for dose values within fibrosis regions. Based on a mixture-Gaussian assumption, an Expectation-Maximization (EM) algorithm was used to obtain characteristic parameters for such distribution. Geometric analysis was performed to interpret such parameters and infer the dose level that is potentially inductive of post-SBRT fibrosis.
Results: The Gaussian mixture obtained from the EM algorithm closely approximates the empirical dose histogram within the fibrosis volume with good consistency. The Kolmogorov-Smirnov test yields a goodness of fit value 0.039. The higher Gaussian component, contributed by the dose received by PTV, was located around the prescription dose (50 or 54 Gy), as expected.
Geometrical analysis suggests the mean of the lower Gaussian component as a possible indicator for a threshold dose that induces fibrosis after SBRT.
Conclusion: Bimodal behavior was observed in the dose distribution of lung fibrosis volumes after SBRT. Novel statistical and geometrical analysis has shown that the systematically quantified low-dose peak is a good indication of a threshold dose for injury. We seek to further improve the quality of registration to extend this analysis to a larger cohort for validation, and to explore planning strategies to reduce post-SBRT injury by controlling the exposure above the threshold value.