A New Independent Outcome Predictor for Lung Stereotactic Body Radiation Therapy Patients, Plan Averaged Beam Modulation
H Liu*, C Li, Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education/Beijing), Peking University Cancer Hospital & Institute, Beijing, CN
Presentations
WE-C1030-IePD-F3-2 (Wednesday, 7/13/2022) 10:30 AM - 11:00 AM [Eastern Time (GMT-4)]
Exhibit Hall | Forum 3
Purpose: Respiratory motion and high dose rate using flattening filter free (FFF) beams for lung stereotactic body radiation therapy (SBRT) may increase MLC interplay effect and delivery uncertainties especially for complex plans involving small fields. However, potential impact of plan complexity on the patient outcomes was not proved by clinical data, which is the purpose of this study.
Methods: Table 1 shows clinical details of 132 plans from 122 patients. High-complexity (HC) and low-complexity (LC) groups were statistically stratified according to plan averaged beam modulation (PM) and plan averaged beam irregularity (PI). PM and PI were defined in Figure 1: Eq.(1-2) and Eq.(3-5) respectively. The cutoffs were calculated using R-3.6.1 package based on the significance of correlation with survival time. Patient outcomes were evaluated using local recurrence-free survival (LRFS). Propensity-score-matched (PSM) pairs were generated to reduce bias. Random Survival Forest (RSF) was used to evaluate the importance of plan complexity on survival predictions. Where MUᵢⱼ is the monitor units (MU) of control point j in beam i; AAᵢⱼ is the area of all MLC openings at each control point; U(AAᵢⱼ) is the union area of all apertures of beam i; AP is the perimeter of MLC aperture.
Results: The median follow-up time was 21.6 months (interquartile range: 11.6-34.5). Cohort PM and PI varied largely ranging from 0.27 to 0.80 and 14.41 to 470.13 respectively. Prognostic capacity of PM was suggested using RSF based on the Variable importance and Minimal depth methods. The survival curves in Figure 2 show that shorter LRFS was significantly associated with higher PM (p=0.034), but was not significant for higher PI (p=0.14).
Conclusion: Based on clinical data, this work demonstrated PM as an independent predictor of lung SBRT outcomes, which can be used as a new tool for pre-treatment plan evaluation to improve patient prognosis.
Keywords
Stereotactic Radiosurgery, Lung, Statistical Analysis
Taxonomy
TH- Response Assessment: Radiomics/texture/feature-based response assessment
Contact Email