Assessment and management of prostate bed motion in post-prostatectomy image guided radiotherapy

Abstract

Introduction Prostate cancer is the most common cancer in Australian men with a radical prostatectomy the usual treatment given for localised disease. Up to 50% of these tumours will have high risk features on pathology review. Positive margin rates range between 13% and 40% and between 20% and 40% of patients will ultimately experience a biochemical failure. These groups should be considered for adjuvant or salvage post-prostatectomy radiotherapy. Accuracy when delivering post-prostatectomy intensity modulated radiotherapy (IMRT) is crucial, both for tumour coverage and avoidance of organs at risk. Standard post-prostatectomy radiotherapy image guidance uses bony anatomy alignment, but the prostate bed can move independently of bony anatomy. Additionally, the upper and lower portions of the prostate bed are likely to move independently of one another due largely to variations in the filling of the bladder and rectum. Quantification of this movement is important for determining the optimal Planning Target Volume (PTV) expansions, which are the margins of normal tissue irradiated around the tumour bed to accommodate variabilities in movement. There is little information available in the literature on optimal margins in the post-prostatectomy setting as well as the feasibility of using anisotropic (asymmetrical) margins to accommodate the prostate’s probable non-uniform movement pattern. Changing from bony anatomy matching to soft tissue matching is another method to potentially increase treatment accuracy whilst minimising the dose to surrounding tissues. Again there is very little information on the potential benefit of this approach, which requires Cone Beam Computed Tomography (CBCT) imaging on a daily basis. For this to succeed, however, clear guidelines must be given to Radiation Therapists on the treatment machine to ensure they understand pelvic anatomy and post-prostatectomy Clinical Target Volumes (CTV). The aims of this study were: to define and quantify the ability to visualise Regions of Interest (ROI) on planning Computed Tomography (CT) and CBCT scans which will enable an atlas to be developed for Radiation Therapists to use when treating post prostatectomy patients, to quantify the magnitude and direction of prostate bed movement, to determine what degree of bladder or rectum size variation creates the potential for geographic miss, to describe prostate bed sagittal tilt and its effect on coverage of the tumour bed, to determine the optimal PTV expansion for post-prostatectomy IMRT when aligning to bony anatomy or soft tissue on a daily basis, and to determine the optimal image guidance policy for post-prostatectomy IMRT. Methods Ethics approval was obtained from the Hawkesbury Research Ethics Committee of Northern Sydney Central Coast Health (0912-377M), the Northern Sydney Central Coast Health Human Research Ethics Committee (1103-082M) and the University of Sydney Human Research Ethics Committee (13721) (see Appendix A). Defining and visualising ROIs Published CTV guidelines were used to define ROIs to be used in soft tissue matching image guidance. A training atlas was made by a Radiation Therapist and two Radiation Oncologists which documented each of the ROIs in descriptive wording and in planning CT and CBCT illustrations. The planning CT scans (n=23) and CBCT scans (n=105) of 23 post-prostatectomy patients were reviewed. Details on ROI identification were recorded. Quantifying prostate bed motion Eligible patients for this and subsequent sections of the analysis included those: (1) receiving adjuvant or salvage intensity modulated post prostatectomy radiotherapy at the Northern Sydney Cancer Centre (NSCC) between 2009 and 2011, (2) who had pre-treatment CBCT scans performed regularly during treatment, and (3) who had a surgical clip located in both the upper and lower portions of the prostate bed, ideally close to midline. Prostate bed motion was calculated in the upper and lower segments by measuring the position of the surgical clips located close to mid-line relative to bony anatomy in the axial (translational) and sagittal (tilt) directions. The frequency of potential geographic misses was calculated for either 1cm or 0.5cm posterior PTV margins. Determining the effect of bladder or rectum size variation on geographic miss Prostate bed movement was correlated with rectal and bladder filling (defined as changes in the cross sectional diameter in the anterior-posterior direction at defined levels). The number of potential geographic misses caused by bladder and rectum variation was calculated, assuming a uniform CTV to PTV expansion of 1cm, apart from 0.5cm posteriorly. Assessing prostate bed tilt The amount of sagittal tilt in the prostate bed was defined as the angle change between a perpendicular baseline at the level of the inferior clip and two surgical clips, one in the upper and one in the lower prostate bed, as close as possible to mid-line. A potential geographic miss was defined as movement by any clip of more than 1cm in any direction or 0.5cm posteriorly, when aligned to bony anatomy. Variations in bladder and rectum size were correlated with the degree of prostate bed tilt and the rate of potential geographic miss was determined. A possible clinical use of prostate bed tilt was then assessed for different imaging techniques. Determining the optimal PTV expansions when aligning daily to bony anatomy Six different PTV expansions were assessed: three published PTV expansions (0.5cm uniform, 1cm uniform, and 1cm + 0.5cm posterior) and three further anisotropic PTV expansions (NSCC, van Herk, and smaller anisotropic). Each PTV was assessed for volume, amount of bladder and rectum irradiated and the potential for geographic miss. Determining the optimal image guidance policy Each CBCT was rematched using a superior soft tissue and averaged soft tissue match which were compared to standard alignment to bony anatomy. Potential geographic miss was assessed using five different PTVs: three published PTV expansions (0.5cm uniform, 1cm uniform, and 1cm + 0.5cm posterior) and two anisotropic PTV expansions (NSCC, and smaller anisotropic). Results Defining and visualising ROIs Images from twenty three patients were analysed, of whom eighteen had surgical clips in situ and five did not. All ROIs were identifiable for all patients on planning CTs at least 90% of the time, apart from the mesorectal fascia (MF) (87%), due to the superior image quality of the diagnostic CT at simulation compared to CBCT images. When using CBCTs in the presence of surgical clips the seminal vesicle bed (SVB) was only seen in 2.3% of images and the MF was unidentifiable, due to the artifact caused by the surgical clips in CBCT imaging. Most other structures were well identified on CBCT. The anterior rectal wall (ARW) was identified in 81.4% of images and the penile bulb in 68.6%. In the absence of surgical clips, the MF and SVB were always identified; the ARW was identified in 89.5% of CBCTs and the penile bulb in 73.7%. An atlas describing these structures on CT and CBCT was developed. The educational value of the atlas was tested in a separate study (Sahota, Bell, Cox, & Atyeo, 2013) and it was found to significantly improve the Radiation Therapists’ structure identification skills. Quantifying prostate bed motion Forty patients met eligibility criteria, incorporating a total of 377 CBCT images. The mean magnitudes of movement of the prostate bed in the anterior-posterior, superior-inferior and left-right directions were: upper portion: 0.50cm, 0.28cm, and 0.10cm and lower portion: 0.18cm, 0.18cm, and 0.08cm. The random and systematic errors of prostate bed motion in the anterior-posterior, superior-inferior, and left-right directions were: upper portion: 0.47cm & 0.50cm, 0.28cm & 0.27cm, and 0.11cm & 0.11cm, and lower portion: 0.17cm & 0.18cm, 0.17cm & 0.19cm, and 0.08cm & 0.10cm. Most geographic misses occurred in the upper prostate bed in the anterior-posterior direction. Determining the effect of bladder or rectum size variation on geographic miss Variations in bladder filling of >2cm larger, ±1cm, or >2cm smaller occurred in 3.4%, 56.2%, and 15.1% of images respectively, with potential geographic misses in the upper prostate bed of 61.5%, 9.9% and 26.3% respectively. Variations in rectal filling in the upper prostate bed of >1.5cm larger, ±1cm, and >1cm smaller occurred in 17.2%, 75.6%, and 7.2% of images respectively. These variations resulted in geographic misses in the upper prostate bed in 29.2%, 12.3%, and 63.0% of images respectively. Variations in bladder and rectal filling in the lower prostate bed region had minimal impact on geographic misses. Assessing prostate bed tilt The median (range) of prostate bed tilt was 1.8º (-23.4º to 42.3º). A tilt of more than 10º, resulting in a 57.9% geographic miss rate of the superior clip, occurred in 20.2% of images. When tilt remained within 10º, there was only a 9% rate of geographic miss. Potential geographic miss of the inferior surgical clip was rare, occurring in only 1.9% of all images reviewed. The most common occurrence when the prostate bed tilt increased by >10º was a smaller bladder and larger rectum (6.4% of all images). The most common occurrence when the prostate bed tilt decreased by >10º was a larger bladder and smaller rectum (1.3% of all images). Determining the optimal PTV expansions when aligning daily to bony anatomy The 0.5cm uniform expansion volume (median = 222.3cc) was the smallest, followed by smaller anisotropic (238.9cc), NSCC (331.5cc), 1cm + 0.5cm posterior (337.7cc), van Herk (354.1cc), and 1cm uniform (361.7cc). Note that the van Herk margin recipe is designed to give 90% of patients at least 95% of the prescribed dose; therefore geographic miss will not be eliminated using this recipe. The van Herk PTV expansion had the smallest rate of geographic miss (4.2%), followed by the 1cm uniform margin (8.0%), NSCC PTV (9.3%), 1cm+0.5 cm posterior margin (15.6%), smaller anisotropic margin (21.0%) and the 0.5cm uniform margin (28.4%). The Van Herk expansion, however, includes the largest amount of bladder (28.0%) and rectum (36.0%), followed by the NSCC (26.4% bladder and 20.0% rectum), 1cm uniform (25.8% bladder; 25.0% rectum), 1cm +0.5 posterior (25.7% bladder; 15.0% rectum), smaller anisotropic (19.3% bladder; 16.6% rectum), and the 0.5cm uniform expansion (17.1% bladder; 10.2% rectum). Determining the optimal image guidance policy Bony anatomy matching resulted in the highest geographic miss rate for all PTVs, followed by superior soft tissue and averaged soft tissue matching. Changing from bony anatomy to an averaged soft tissue match decreased potential geographic miss by half to two thirds, depending on the PTV expansion. The averaged soft tissue matching technique reduced geographic miss rates to <10% for all PTV expansions. When using the smaller anisotropic PTV expansion, the use of averaged soft tissue matching would reduce the geographic miss rate from 21.0% with bony anatomy matching to 5.6%. Conclusions Most relevant pelvic structures are identifiable on CBCT images, enabling their use for online matching. Surgical clips should be used as ROIs when present to define SVB and MF. In the absence of clips, SVB, MF and ARW can be used. The prostate bed moved independently of the bony anatomy with variability seen in all directions for both superior and inferior surgical clips. Greatest movement occurred in the anterior-posterior direction in the upper prostate bed, raising concerns for potential geographic miss during treatment delivery with conventional margins and image verification. The variability in movement between the superior and inferior clips results in a prostate bed tilt that would be difficult to correct with standard on-line matching techniques. This creates a strong argument for using anisotropic PTV margins in post-prostatectomy radiotherapy. Bladder and rectal size changes at treatment affect prostate bed coverage, especially in its upper aspect. Considerable prostate bed sagittal tilt (>±10º) occurred in more than 20% of images, creating a 58% rate of geographic miss. Greatest prostate bed tilt occurred when the bladder size increased or reduced by more than 2cm or the superior rectum size increased by >1.5cm or reduced by >1cm from the planned size. Ensuring a full bladder and empty rectum at simulation will help minimize this risk. Measurement of prostate bed tilt could be an effective tool for assessing potential geographic miss on orthogonal images if volumetric imaging is unavailable. Anisotropic PTV expansions generated the fewest geographic misses when bony anatomy matching was used. The van Herk anisotropic margins had the lowest geographic miss (4.2%), but resulted in unacceptable bladder and rectal overlap. The NSCC anisotropic margin provides the best balance between geographic miss (9.3%) and minimising bladder and rectal volumes within the PTV. The optimal image guidance policy for post-prostatectomy radiotherapy is daily average soft tissue matching using CBCT scans with an anisotropic PTV expansion. Daily averaged soft tissue matching enables consideration of reduced PTV expansions and adaptive radiotherapy. Caution must still be taken during online matching to ensure surrounding critical structures are not receiving increased dose. There are a number of limitations of this study, with the main limitation being that it is a single observer, single institution study. Further research needs to be made into the management of motion in the post-prostatectomy setting, which could include adaptive radiotherapy to reduce margins further.