Improving plan quality for prostate volumetric-modulated arc therapy.

We critically evaluated the quality and consistency of volumetric-modulated arc therapy (VMAT) prostate planning at a single institution to quantify objective measures for plan quality and establish clear guidelines for plan evaluation and quality assurance. A retrospective analysis was conducted on 34 plans generated on the Pinnacle3 version 9.4 and 9.8 treatment planning system to deliver 78 Gy in 39 fractions to the prostate only using VMAT. Data were collected on contoured structure volumes, overlaps and expansions, planning target volume (PTV) and organs at risk volumes and relationship, dose volume histogram, plan conformity, plan homogeneity, low-dose wash, and beam parameters. Standard descriptive statistics were used to describe the data. Despite a standardized planning protocol, we found variability was present in all steps of the planning process. Deviations from protocol contours by radiation oncologists and radiation therapists occurred in 12% and 50% of cases, respectively, and the number of optimization parameters ranged from 12 to 27 (median 17). This contributed to conflicts within the optimization process reflected by the mean composite objective value of 0.07 (range 0.01 to 0.44). Methods used to control low-intermediate dose wash were inconsistent. At the PTV rectum interface, the dose-gradient distance from the 74.1 Gy to 40 Gy isodose ranged from 0.6 cm to 2.0 cm (median 1.0 cm). Increasing collimator angle was associated with a decrease in monitor units and a single full 6 MV arc was sufficient for the majority of plans. A significant relationship was found between clinical target volume-rectum distance and rectal tolerances achieved. A linear relationship was determined between the PTV volume and volume of 40 Gy isodose. Objective values and composite objective values were useful in determining plan quality. Anatomic geometry and overlap of structures has a measurable impact on the plan quality achieved for prostate patients being treated with VMAT. By evaluating multiple planning variables, we have been able to determine important factors influencing plan quality and develop predictive models for quality metrics that have been incorporated into our new protocol and will be tested and refined in future studies.