Investigation of uncertainties associated with the MammoSite™ Breast Brachytherapy Technique: Monte Carlo simulations and TLD measurements

Abstract

The MammoSite Radiation Therapy System is a novel brachytherapy technique for treatment of patients with early stage breast cancer. It is used as a sole radiation treatment or in combination with external beam radiotherapy. There are several uncertainties associated with the dose distribution from the MammoSite brachytherapy. In this research study, the ¹⁹²Ir brachytherapy source was accurately modelled using the EGSnrc Monte Carlo code. A voxel size of 1.5 mm³ was found to be suitable for dose calculations as reducing the voxel size any further would increase the simulation time without improving the accuracy of dose simulation. The impact of uncertainties in balloon deformation and source position on the tumour control probability (TCP) and the normal tissue complication probability (NTCP) were assessed. The effects on the treatment outcome were assessed from (a) organ differential dose volume histograms (dDVHs) obtained from the Plato brachytherapy planning system and (b) EGSnrc Monte Carlo simulations based on actual computed tomography (CT) images of a breast cancer patient who underwent MammoSite brachytherapy treatment. This study gave low probabilities for developing heart and lung complications. Monte Carlo calculations showed that a deviation of the source by 1 mm caused approximately 7% dose reduction in the treated target volume at 1 cm from the balloon surface. A 4 mm source deviation produced underdosing of some portions of the PTV by 40% leading to poor treatment outcomes. Furthermore, 4 mm uncertainty in source deviation leads to overdosing of regions of the PTV by about 40%. This results to an excessive dose to the skin and increases the probability of skin complications. Balloon deformation and incorrect source position had significant effect on the prescribed dose within the treated volume. A 4 mm balloon deformation resulted in reduction of the tumour control probability by 24%. The current study suggested that the MammoSite treatment protocols should allow for a balloon deformation of less than 2 mm and a maximum source deviation of ≤ 1 mm. The extent of the dose perturbation for various concentrations of the contrast medium in a MammoSite balloon was investigated using Monte Carlo simulations and thermoluminescent dosimeters (TLDs) measurements. The Monte Carlo simulation was performed using CT images of in-house tissue equivalent breast phantom. The breast phantom was also used for TLD measurements. The measured and Monte Carlo calculated doses were in agreement within the measurement uncertainty and Monte Carlo statistical errors. The dose reduction resulting from the use of high atomic number contrast (Iodine) caused considerable uncertainty in the MammoSite dose. Our results showed that 100%, 50% and 15% contrast concentrations reduced the dose at the prescription point by 10%, 5% and 2% respectively relative to the dose calculated with the balloon filled with saline (water) only. The BEAMnrc and DOSXYZnrc Monte Carlo codes were used to model an external beam radiotherapy treatment and simulate a dose distribution using a patient CT data set respectively. The external beam radiotherapy model was validated with measurements and the data analysis was performed using the gamma function algorithm. The gamma function analysis algorithm was used and the acceptance criteria for comparison were set to distance-to-agreement of 2 mm and 2% dose difference. An excellent agreement (99.4% of detectors passed the criteria) was found between the Monte Caro computed dose maps and the measured ones. This proved that a reliable Monte Carlo model was constructed and used for dose calculations from EBRT treatment. The simulated dose distribution from EBRT was combined with the simulated MammoSite dose distribution. Finally, it would be beneficial (to the oncologist) to visualize the final (combined) dose distributions from the two modalities to assist with an assessment of treatment plans and the treatment outcome. Currently, combining the dose distributions from the two modalities is difficult to achieve because the two modalities use different planning systems and different dose calculations algorithms and the patient anatomy looks different (balloon is present for brachytherapy). Consequently, the project aimed to build a Monte Carlo linac model to calculate dose delivery to a breast due to external beam radiotherapy. Having MC models (EBRT & MB) constructed and verified, the dose distributions calculated from each modality were converted using appropriate algorithms to equivalent dose distributions and combined to yield the total dose distribution to a breast from the combined treatment.