RBE Model Based Proton Planning of Stereotactic Body Radiotherapy for Spine Metastasis: A Dosimetric Analysis

Abstract

Study of proton stereotactic body radiotherapy (SBRT) for spine metastasis is limited, largely due to theorized increased risk of spinal cord injury with higher end of range RBE. Though the 1.1 RBE constant for proton beam is clinically used, data indicate that proton RBE is variable and dependent on technical-, tissue-, and patient-factors. To better understand safety of proton SBRT for spine metastasis, we performed a dosimetric analysis comparing plans delivered by photon robotic technique versus intensity modulated proton therapy (IMPT) and accounting for RBE weighted dose. A total of 9 patients with spine metastasis (3 cervical, 3 thoracic, 3 lumbar) previously treated with a frameless robotic radiosurgery system (Sunnyvale, CA) were identified. Each level contained a case with paraspinal extension, a reirradiation case, and a case with epidural extension (Bilsky grade ≥1c) as such cases in current practice often require planning target volume (PTV) under-coverage in order to meet organ at risk (OAR) dose constraints. Given these challenges, selected cases were clinically treated with 30 Gy in 5 fractions despite an institutional preference of further dose escalation. Comparative IMPT plans were generated using 30 GyE in 5 fractions and 1.1 RBE constant. IMPT plans were then made using 1.1 RBE and 45 GyE in 5 fractions: a prescription dose associated with a 2-yr local control rate of 95% on prior tumor control probability modelling. A treatment planning system was used to separately generate and optimize RBE weighted plans based on Carabe-, McNamara-, or Wedenberg models for prescription doses of 30 GyE and 45 GyE. IMPT plans used robust optimization parameters of ± 3.5% range and 2-mm setup uncertainties. PTV coverage and OAR sparing were compared using Wilcoxon signed-rank tests. PTV coverage (PTV volume receiving prescription dose) was significantly improved with IMPT at 30 GyE / 1.1 RBE (median PTV V30: 93%) compared to CK at 30 Gy (median: 88.5%, p = .02). PTV coverage was similar when comparing CK at 30 Gy with IMPT at 45 GyE / 1.1 RBE (median PTV V45: 90%, p = .23). When comparing maximum spinal cord dose (cord Dmax), there was improved OAR sparing with IMPT at 30 GyE / 1.1 RBE (median: 17.6 GyE, p = .04) and IMPT at 45 GyE / 1.1 RBE (median: 16.1 GyE, p = .04) when respectively compared to CK at 30 Gy (median: 18 Gy). No difference was seen in cord Dmax when comparing CK at 30 Gy to RBE weighted plans at 30 GyE using Carabe- (median: 17.3 GyE, p = .22), McNamara- (median: 17.4 GyE, p = .22), or Wedenberg (median: 17.0 GyE, p = .08) model. Median cord Dmax values for RBE weighted plans at 45 GyE were numerically equivalent. The average increase in variable RBE plans' maximum dose compared to fixed RBE plans was 105.3% +/- 3.5%. We report the first dosimetric analysis of proton SBRT for spine metastasis using variable RBE dose models. IMPT may provide improved target coverage and better sparing of adjacent OARs compared to CK though fixed RBE computation may underestimate maximum dose to adjacent OARs.