Search

Oesophageal cancer continues to be associated with a poor prognosis. Proton beam therapy's distinct physical characteristics widen the therapeutic ratio in oesophageal cancer and has the potential to improve outcomes. This thesis aims to examine how proton beam therapy may improve outcomes in oesophageal cancer and documents efforts to expand its role. In Chapter 2, a comprehensive systematic literature review demonstrates the paucity of high-quality evidence in this field. Following this, a series of radiotherapy planning studies investigates potential dosimetric advantages of proton beam therapy for distal oesophageal cases. In chapter 3, proton beam therapy is shown to reduce lung and heart dose compared to photon radiotherapy. Normal tissue complication probability modelling establishes that this may reduce the risk of treatment related pulmonary and cardiac toxicity. Chapter 4 demonstrates that spleen dose constraints may successfully be introduced in oesophageal cancer for proton and photon plans, potentially resulting in lower lymphopenia rates and greater immune sparing. Chapter 5 highlights the impact of different beam arrangements on dose to organs at risk and individual cardiac substructures. The latter half of the thesis highlights work underpinning the development of novel clinical trials of proton beam therapy in oesophageal cancer. Chapter 6 details the work of creating a radiotherapy delineation protocol by comparing two established protocols in a delineation comparison study, showing that geometric expansion of volumes results in more consistent target volumes compared to 'free-hand' delineation. In chapter 7, public and patient involvement work is shown to inform and refine the design of a two new trials of proton beam therapy in oesophageal cancer. A final chapter discusses and summarises the current areas of interest in this field, expanding on current trial development work and future directions.

Proton therapy is an emerging modality for providing radiation treatment to cancer patients. The principal advantages of proton therapy are the reduced total dose deposited into the patient as compared to conventional photon therapy and the finite range of the proton beam. It is considered as a more favorable option for optimum treatment outcomes in terms of maximising tumor control probability and minimising normal tissue complications. The depth dose distribution of the proton beam adds an additional degree of freedom to treatment planning. The range in tissue is associated with substantial uncertainties triggered by imaging, patient set-up, beam delivery and dose calculations. Therefore, reduction in uncertainties would allow to minimise the treatment volume and thus allow a better usage of the protons. However, the presence of sub-millimeter sized heterogeneities, such is pronounced in trabecular bone and lung parenchyma, in the path of the proton beam can cause the Bragg peak degradation with a widening to the distal fall-off. Additionally, the restricted resolution of a classic CT scan used in treatment planning cannot fully resolve such fine structures, potentially leading to inaccuracy in determination of the range. This work aims to investigate the presence of range uncertainties in proton therapy beams when they penetrate through the sub-millimeter sized heterogeneities. The effect of Bragg peak degradation has been demonstrated in bone models with the FLUKA Monte Carlo code and experimental measurements with a 36 MeV proton beam. The bone-substitute material, SAWBONES®, ranging in density from 0.088 to 0.48 g/cc, was used to simulate bone heterogeneities. Micro-CT images were obtained of the SAWBONES® material and used to construct Monte Carlo models of realistic proton radiotherapy treatments and to benchmark experimental studies. Broadening of the Bragg peak and shifts in the range, as defined by the d20% depth-dose parameter were observed both experimentally and in Monte Carlo models, indicating that such effects are in principle, clinically relevant in certain circumstances. Furthermore, a FLUKA Monte Carlo model is benchmarked against the Eclipse treatment planning system (TPS) golden data for proton beam therapy. This project is designed to obtain the proton dose distributions from TPS for a 10 ×10 × 10 cm3 water-filled box. A Monte Carlo analytical model is developed by utilising the information from the TPS to recalculate the dose distribution which are then compared to find any differences (if present) for different phantom materials. Due to the lack of any experimental information to measure the normalized depth dose as a function of energy, considering the general behaviour of a monitor chamber, it has been assumed that the treatment planning system has a built-in relationship between the monitor units (MU) and dose delivered. A mathematical formula is developed to find the relationship between monitor units (MU) and dose (E). MU α 1/Ea. The value of a is varied from 0 to 1. It has been observed that the beam non-uniformity calculated by using the relationship "MU = E−0.5" is only 0.15 % for water and 0.43 % for graphite. This non-uniformity in graphite is not severe and it is actually clinically acceptable. This modeling has suggested that the planned dose distributions for water can also be replaced by graphite to a reasonably acceptable standard.

Microbeam Radiation Therapy (MRT) is an emergent treatment modality that uses spatially fractionated synchrotron X-ray beams. MRT has been identified as a promising treatment concept that might be applied to patients with malignant central nervous system (CNS) tumors for whom, at the current stage of development, no satisfactory therapy is available yet. The use of a fractionated beam allows a better skin sparing and a better tolerance of healthy tissue to high dose rates. MRT consists of a stereotactic irradiation with highly collimated, quasi-parallel array of narrow beams 50 μm wide spaced with 400 μm made of synchrotron generated X-rays at an energy ranging from 0 to 600 keV. The European Synchrotron Radiation Facility (ESRF) as an X-ray source allows a very small beam divergence and an extremely high dose rate. The dose deposited on the path of the primary photons (peak dose) of several hundred grays (Gy) is well tolerated by normal tissues and provides at the same time a higher therapeutic index for various tumor models in rodents. The high dose rate forces us to develop an accurate and reproducible dosimetry protocol to ensure the matching between the prescribed and the delivered dose. MRT is by definition a non-conventional irradiation method, therefore the number of dosimetric errors becomes larger than in conventional treatments due to two reasons (i) the reference conditions recommended by the Association of Physicists in Medicine (AAPM) or the International Atomic Energy Agency (IAEA) cannot be established, (ii) the measurement of absorbed dose to water in composite fields is not standardized. This PhD is focused on bridging the gap between MC simulated values of output factors (OF) and peak-to-valley dose ratios (PVDR) and experimental measurements. Several aspects of the irradiation setup such as insertion devices on the path of the X-ray beam are accounted for as well as the internal structure of the dosimeters. Each contribution to OF and PVDR is quantified to correct for the measurements.

Objective: Cone beam CT (CBCT) images contain more scatter than a conventional CT image and therefore provide inaccurate Hounsfield units (HU). Consequently CBCT images cannot be used directly for radiotherapy dose calculation. The aim of this study is to enable dose calculations to be performed with the use of CBCT images taken during radiotherapy and evaluate the necessity of re-planning. Methods: A phantom, a standard prostate cancer patient and prostate cancer patients with single and double metallic hips were imaged using both CT and CBCT. A multilevel threshold algorithm (MLT) was used to categorise pixel values in the CBCT images into segments of homogeneous HU. The variation in HU with position in the CBCT images was taken into consideration and the benefit of using a large number of materials has been explored. This segmentation method relies upon the operator dividing the CBCT data into a set of volumes where the variation in the relationship between pixel values and HUs is small. In addition, an automated MLT algorithm was developed to reduce the operator time associated with the process. Furthermore, magnetic resonance (MR) images of the standard prostate case were segmented and converted into HUs using the MLT algorithm. Radiotherapy treatment plans were generated from CT images and then copied to the segmented CBCT and MR data sets and the doses were recalculated and compared using pencil beam (PB), collapsed cone (CC) and Monte Carlo (MC) algorithms. Results: Compared with the planning CT (pCT) treatment plan, in the phantom case, a gamma evaluation showed all points in planning target volume (PTV), rectum and bladder had gamma value < 1 (3%/3 mm) in the segmented CBCT, when considering only 2 material bins, water and bone. For the standard patient case, using 3 materials, air, water and bone, was accurate enough to provide accurate dose calculations with differences of less than 2%. For the patient with a metallic hip, increasing the number of bins to define the material type from 7 materials to 8 materials, required 50% more operator time to improve the accuracy by 0.01% using PB and CC and 0.05% when using MC algorithms. The use of 5 values of HU (air, adipose, water, bone and metal implant) gave the best balance between dose accuracy and operator time (3.5 hours). For the patient with double hip prosthetics, segmenting CBCT into 5 materials with the MLT algorithm showed -0.46% dose difference with 8 hours operator time, whilst the automated MLT algorithm showed -1.36%. For the standard case, the segmentation of MR images, into 3 materials, resulted in a dose difference of -1.31% with 2 hours operator time. Conclusion: The segmentation of CBCT images using the method in this study can be used for dose calculation. For a simple phantom and standard prostate case, 2 and 3 values of HU were needed to improve dose calculation accuracy, respectively. For patients with additional anatomical inhomogeneities such as metallic hips, 5 values of HU were found to be needed, giving a reasonable balance between dose accuracy and operator time. The automated MLT algorithm reduced the operator time associated with implementing the MLT algorithm to achieve clinically acceptable accuracy. This saved time makes the automated MLT algorithm superior and easier to implement in the clinical setting. The MLT method can be applicable for the dose calculation on MR images and can be of interest to MRI-only based radiotherapy treatment planning.