Effect of imaging frequency on PTV margins and geographical miss during image guided radiation therapy for prostate cancer.

BACKGROUND

The relationship between frequency of imaging during image guided radiation therapy (IGRT) and planning target volume (PTV) margin remains unclear. This issue is of practical significance given resource and time intensive nature of IGRT. The purpose of this study was to evaluate PTV margins with predefined and commonly used less-than-daily IGRT schedules using data obtained from patients treated with daily IGRT for prostate cancer.

METHODS AND MATERIALS

Daily setup error and 3-dimensional daily alignment data for a total of 108 consecutive patients with prostate cancer treated with 2700 fractions of daily image guidance on tomotherapy were retrospectively analyzed. Five IGRT scenarios were simulated: alternate day, twice weekly, once weekly, first 3 days only, and no image guidance. The daily alignment data were modeled to simulate the 5 predefined scenarios by applying appropriate corrections to determine the PTV margin for each image guidance scenario. The data were also analyzed to predict possible geographical miss in any direction using 2 frequently used PTV margins of 7 and 5 mm for all the scenarios.

RESULTS

Decreasing the frequency of image guidance increased the mean systematic error and the standard deviation of the systematic error. With decreased image guidance frequency, an increase in PTV margins was required to achieve adequate coverage of the clinical target volume. With reduction in image guidance from 50% to 12%, a gradual increase in percentage of fractions with predicted geographical miss using an isotropic PTV margin of 7 or 5 mm was seen. With every 15% decrease in imaging, a 5% increased risk of geographical miss was estimated.

CONCLUSIONS

The use of less-than-daily IGRT requires larger PTV margins for patients treated with intensity modulated radiation therapy for prostate cancer. With every 15% reduction, a 5% increased risk of geographical miss was estimated.