Program Information
Single-Isocenter Multiple-Target SRS: Risk of Compromised Coverage
J Roper1,3*, V Chanyavanich1,3 , G Betzel1,3 , J Switchenko2,3 , A Dhabaan1,3 , (1) Department of Radiation Oncology, (2) Department of Biostastics and Bioinformatics, (3) Winship Cancer Institute, Emory University, Atlanta, GA
Presentations
TH-A-9A-11 Thursday 7:30AM - 9:30AM Room: 9APurpose: To characterize the risks of compromised coverage in single-isocenter multiple-lesion VMAT SRS.
Methods: Fifty patients were selected retrospectively from our SRS program. Each patient had two lesions treated to ≥ 21 Gy. Single-isocenter VMAT SRS plans were created in Eclipse. PTV volumes and distances from isocenter were recorded. PTV coverage (D95 and V95) was evaluated across rotational setup errors of 0.5, 1.0, or 2° applied to three axes. Coverage rates were analyzed versus volume, distance, and rotation. For a rotational error of 2°, lesion size and separation distance were compared across coverage rate levels using ANOVA. A multivariate logistic regression model was fit using generalized estimating equations (GEE), modeling the probability of a 95% V95/D95 rate or higher given lesion size and separation distance while accounting for intra-patient correlation. The estimated probabilities from the GEE model were used to capture the operating conditions in a receiver operating characteristic (ROC) curve; area under the curve (AUC) was estimated.
Results: Mean lesion volume and distance to isocenter are 0.96±1.25cc and 3.53±1.61cm. V95/D95 proportions above 95% range from 92-100% when rotational errors are ≤1°. At 2.0° rotation, V95/D95 are >95% in only 62-64% of cases; V95 falls to 75% for <0.3cc lesions at 4cm yet remains >90% up to 6cm for lesions >0.9cc. V95 is <40% in an extreme case. The logistic regression analysis shows that lesion volume and distance to isocenter are independent predictors (p< 0.001) of V95/D95 rates exceeding 95%. The ROC derived from a GEE multivariate model has an AUC of 0.87.
Conclusion: PTV coverage can be compromised substantially by rotational setup errors of 2°, in particular for <0.3cc lesions at distances >4cm from isocenter. Statistical analysis suggests that lesion volume and distance to isocenter could be used to select patients who are good candidates for single-isocenter multiple-lesion SRS.
Contact Email: