Evaluation of the Impact of the Linac MLC and Gantry Sag in volumetric modulated arc therapy

Purpose Mechanical sag in the radiotherapy linear accelerator gantry and multi‐leaf collimator (MLC) carriage effectively causes systematic deviations in the isocenter with respect to gantry angle. To minimize the impact of this error on treatment, a tolerance value of a 1‐mm mechanical isocenter shift is commonly accepted for intensity‐modulated radiation therapy quality assurance (QA). However, this tolerance value has not been firmly established for volumetric modulated arc therapy (VMAT) treatments. The purpose of this study is therefore to evaluate the impact of gantry and MLC carriage sag on VMAT clinical performance. Methods A published dataset of Elekta and Varian sag measurements served as a starting point for the investigation. Typical sag profiles were chosen and modeled as continuous isocenter deviations in three dimensions. The data were then incorporated into existing Digital Imaging and Communications in Medicine protocol, extended for radiotherapy plans via a “beam‐splitting” algorithm. Three treatment sites were investigated in parallel: head and neck, prostate, and prostate with surrounding lymph nodes. Monte Carlo‐simulated dose distributions were obtained for varying magnifications of the modeled sag. The resulting dose distributions, including that for no error, were compared qualitatively and quantitatively, against multiple metrics. Results The dose‐volume histograms (DVHs) for all plans exhibited a decrease in planning target volume (PTV) dose uniformity with increasing sag magnification, whereas dose to organs at risk exhibited no coherent trend. The prostate plan was shown to be the most vulnerable to mechanical sag across all considered metrics. However, all plans with peak isocenter deviation less than 1 mm were well within typical cutoff points for each metric. Conclusions All avenues of investigation presented substantiate the commonly accepted tolerance value of a 1‐mm peak isocenter shift in annual linac QA.