Purpose: The link between radiotherapy bladder dose and chronic urinary toxicity is unclear. This thesis took the unique stance that robust associations between bladder dose and toxicity will not be found unless the dose incorporates the effects of daily bladder motion. The purpose of this thesis was to develop and validate novel techniques to reconstruct the dose delivered to the normal bladder functional tissue during fractionated radiotherapy. Methods: This was a 7-phase quantitative research design. Phase-1 evaluated bladder and trigone delineation on planning CTs (planCT) for 30 patients. Phase-2 evaluated bladder wall delineation on treatment images (CBCT) for 10 patients. Phase-3 varied the calculation dose grid spacing for 15 patients to determine the optimal interval for bladder wall (BW) and trigone DVHs. Phase-4 evaluated the use of linear elastic modelling and finite element analysis (LEM/FEA) to deform the bladder surfaces and intra-wall elements for 24 patients. Phase-5 determined whether LEM/FEA software could accurately render the initial dose distribution and then map elements within that distribution for 10 patients. Phase-6 combined techniques developed in the previous phases to reconstruct bladder and trigone delivered dose for 30 patients, exploring factors that influenced the difference between planned dose (planDose) and reconstructed dose (reconDose). Phase-7 associated bladder and trigone reconDose with an objective indicator of chronic urinary toxicity for 59 patients. Results: Semi-automated bladder delineation differed <2cm3 from manual delineation on planCT, but demonstrated less observer variability (Intraclass Correlation Coefficient [ICC] >0.77 vs. >0.49; p 0.001). Manual trigone delineation was associated with substantial observer variability (ICC <0.49), but less variable automated methods failed to accurately approximate the dose (-20%; p 0.001). Propagated CBCT contours were slightly larger than manual contours (+8%; p 0.001), but demonstrated less observer variability (Dice Similarity Coefficient 0.84 vs. 0.90; p 0.008). The use of an equation to delineate IB on CBCT differed <2cm3 from manual delineation. Dose grid spacing >1.5mm significantly underestimated bladder volume receiving 78Gy (-3cm3 ; p 0.002). LEM/FEA transformed surface and intra-wall element positions with an accuracy of <1mm, and rendered and mapped the dose to within 0.5% of the treatment planning system. Bladder reconDose was 8Gy higher than planDose for prostate IMRT and cervix 4-field, but lower for rectum 3-field (p <0.05). Trigone reconDose was similar to planDose for cervix 4-field (p >0.1), but higher for prostate IMRT and lower for rectum 3-field (+6Gy and -9Gy, respectively; p <0.016). The difference between bladder planDose and reconDose was dependent on bladder filling variation and plan dose gradient, and was effected by the number and sequence of images included in the reconstruction process. The incidence of chronic urinary toxicity was associated with the bladder volume within the reconstructed 35-50Gy isodose region (1cm3 increment: Odds Ratio [OR] 1.06; p 0.03). Bladder planDose was correlated with reconDose in that region (R2 >0.5), and was also associated with toxicity (OR 1.07; p 0.03). Neither trigone planDose nor reconDose was related to toxicity (OR 0.79-1.2; p 0.77). Conclusions: The methods developed and tested in this thesis were able to accurately and reproducibly reconstruct the dose delivered to the bladder functional tissue, such that clear associations between bladder dose and an objective indicator of urinary dysfunction were identified. Using these methods to reconstruct bladder delivered dose has the potential to reveal the relationship between bladder dose and chronic toxicity within a prospective clinical trial, and contribute to minimizing the risk of toxicity using inverse planning, improve the accuracy of radiobiological predictions and assist in the identification of useful biomarkers.
|Qualification||Doctor of Philosophy|
|Award date||15 Apr 2015|
|Place of Publication||Australia|
|Publication status||Published - 2015|