Dosimetric Comparison and Clinical Implications of Dose Calculation Algorithms for Stereotactic Body Radiation Therapy in Spine Metastases.

Abstract

PURPOSE/OBJECTIVE(S)\nDose calculation algorithms for radiotherapy differ in their ability to handle patient-specific anatomy, tissue heterogeneity, and artificial implants, which is crucial in spine stereotactic body radiation therapy (SBRT), where a large dose gradient exists between normal organs and the tumor. We evaluate the dosimetric differences in 3 calculation algorithms: Collapsed Cone Convolution (CCC), Anisotropic Analytic Algorithm (AAA), and Acuros XB (AXB) for spine SBRT with and without metal implants. We hypothesize that clinical outcomes can be correlated with the accuracy of the algorithm and will show this through 3 clinical scenarios.\n\n\nMATERIALS/METHODS\nWe retrospectively analyzed 41 spine SBRT plans from 27 patients, 15 with metal hardware. CCC plans (17) were recalculated with AAA and AXB and AXB plans (24) were recalculated with AAA. Planning target volume (PTV) coverage and maximum dose (Dmax) for spinal cord were compared with AXB as the standard, as it is the most accurate in accounting for tissue heterogeneity of the three algorithms. Chart review was completed to determine clinical correlations.\n\n\nRESULTS\nTumors of various histologies including Lung (22.0%), breast (19.5%), head & neck (17.1%), Renal Cell (14.6%) were studied. Treatments were delivered in 1-5 fractions (median 3) with Rx dose of 16-40 Gy (median 27), corresponding to BED10 of 37.5-60 Gy (median 51.3). Overall (see Table), AAA tended to overestimate Dmax while CCC underestimated. Both algorithms overestimated the dose to 95% of PTV (D95). Those with metal near the target had a larger dose discrepancy, especially with CCC. With median follow up of 14.7 mo, poor local control was noted in 14 (34%) plans. Compared to plans with local control, these were more likely to have metal implants near the PTV (50%), with a larger degree of overestimation of PTV coverage compared to AXB, especially in plans calculated using CCC (Avg. 7.3% overestimation of D95).\n\n\nCONCLUSION\nAAA tends to overestimate both Dmax to spinal cord and PTV coverage compared to ABX. CCC tends to underestimate Dmax to the cord while also overestimating coverage. The differences were larger in the presence of metal implants. Larger dose differences were also noted in lesions that showed poor local control, especially with CCC. The clinical use of more accurate calculation algorithms such as AXB may result in improved outcomes of spine SBRT, especially in the post-operative setting.