Role of Residual Setup Errors and Intrafraction Motion in the Evaluation of Planning Target Volume Margins for Prostate Cancer Radiation Therapy

Traditional setup techniques rely on skin marks aligned to room-based lasers or on the alignment of bony structures. However, these techniques do not take into account that the prostate moves relative to both skin marks and bony anatomy. With daily online correction based on implanted fiducial markers as a surrogate for prostate position, a significant improvement of the accuracy of daily prostate localization can be achieved. However, residual set-up errors (RSE) reflecting inaccuracies in the imaging and repositioning systems as well as intrafraction motion may become limiting factors in prostate targeting accuracy. The present study was performed to quantify and characterize interfraction deviations as well as RSE and intrafraction prostate motion assessed by multiple kilovoltage (kV) and megavoltage (MV) imaging of implanted fiducial markers during prostate cancer radiation therapy and to determine PTV margins for different alignment techniques. The present study comprised 44 prostate cancer patients who underwent a three-field 3D conformal treatment. In all patients, four commercially available gold markers were implanted into the prostate transrectally. Daily localization was based on skin mark alignment followed by marker detection in kV- kV imaging and patient repositioning. Furthermore, in-treatment MV images were obtained for each treatment field. In an off-line analysis of all kV and MV images, the RSE after marker-based prostate localization, the prostate position during a radiation therapy session and the effect of treatment time on intrafraction deviation were evaluated. Mean RSE and standard deviation (SD) in the antero-posterior (AP), superior-inferior (SI) and left-right (LR) direction were 0.5 ± 1.1 mm, 0.2 ± 0.6 mm and 0.2 ± 0.6 mm, residual rotational deviations around the LR and AP axes were 3.6 ± 3.4 deg and 0.9 ± 1.0 deg, respectively. Mean intrafraction motion in AP, SI, and LR direction was 0.7 ± 1.7mm, 0.8 ± 1.4mm, 0.3 ± 0.8 mm, rotational movements around the LR and AP axes were 1.7 ± 3.5 deg and 0.4 ± 1.1 deg. Intrafraction motion increased with prolongation of treatment. Considering positioning errors only, the required PTV margins were for skin mark alignment 14 mm (AP), 13 mm (SI), and 15 mm (LR), bony structure alignment 10 mm (AP), 9 mm (SI), and 3 mm (LR) and alignment to implanted markers 5 mm (AP), 3 mm (SI), and 3 mm (LR). PTV margins can further be reduced to 4 mm (AP), 2.4 mm (SI), and 2.6 mm (LR) if treatment time is ≤ 4mm. With daily online correction and repositioning based on implanted markers, a significant reduction of PTV margins can be achieved. As the margins required for intrafraction motion increased with prolongation of treatment, the use of an optimized workflow with faster treatment techniques could allow for a further decrease in the PTV margin.