Graduation Year

2024

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Physics

Major Professor

Gage Redler, Ph.D.

Co-Major Professor

Ghanim Ullah, Ph.D.

Committee Member

Eduardo G. Moros, Ph.D.

Committee Member

Jimmy J. Caudell, MD, Ph.D.

Keywords

head and neck cancer, polymetastatic disease, low dose radiotherapy, adaptive radiation

Abstract

Radiation therapy (RT) is an important pillar of modern cancer treatment, in neoadjuvant, curative, adjuvant, and palliative settings. The physics underlying RT is ionizing radiation that produces charged particles, which deposit their energy in the tissues they pass through. This energy results in unrepairable DNA damage resulting in the death of cancer tissue. The aim of RT is to deliver the desired radiation dose to the tumor while maximizing normal tissue sparing. Over the past decades, physics, engineering, and computational improvements has allowed advancement of external beam RT techniques, for example from 2D RT to 3D RT to now intensity modulated RT (IMRT) and volumetric modulated arc therapy (VMAT). 2D RT uses large RT fields based on clinical information and conventional X-rays. 3D conformal RT uses smaller RT fields that are shaped to the tumor, based on volumetric imaging, resulting in improved outcomes and reduced treatment toxicity. The newer IMRT and VMAT treatment modalities allow additional optimization of complex dosimetric gradients to provide highly conformal dose to the tumor with reduced high to intermediate dose normal tissue exposure. Conventional RT is based on creating a treatment plan with one of these modalities, which will be used for the whole course of treatment that may last from 5 to 7 weeks. To improve the accuracy of utilizing the same plan throughout the course of treatment, imaged guided RT (IGRT) was developed to reduce uncertainties associated with daily treatment set-up. This conventional approach still utilizes additional margins around the target to account for remaining uncertainties in positioning. In reality, due to the treatment course lasting 4-7 weeks, many changes are possible and sometime inevitable. While IGRT can account for rigid positioning changes, more complex deformations such as weight loss, tumor growth/shrinkage and patient movement may not be fully accounted for. Adaptive RT (ART) is based on updating the initial treatment plan based on the new anatomy to account for uncertainties that could results in dosimetric variations. In this work, we aim to investigate and evaluate the Ethos CBCT-guided ring gantry online ART (oART) system, based on the Halcyon machine, with a semi-automated treatment planning and an online adaptive platform that accounts for changes at each individual session during the course of treatment.

In the first chapter, a general overview of conventional RT as well as oART is presented to differentiate between the process and the accuracy/uncertainties of the two treatments. We have also presented the two sites that we have performed our study on: head and neck (H&N) and polymetastatic disease in stage IV non-small cell lung cancer (NSCLC).

In the second chapter, a general overview of the Ethos oART system is presented, with an in depth discussion of the intelligent optimization engine (IOE) and the optimization objectives that are automatically generated to produce dose distributions based on clinical goals provided by the user. We have briefly described the dose optimization process, which is comprised of the photon optimization algorithm (PO) and dose calculation algorithms (Fourier transform dose calculation (FTDC) and Acuros XB) used in the Ethos system.

In the third chapter, we have developed an initial planning strategy for H&N patients treated with sequential boost (SEQ) and simultaneous integrated boost (SIB) in the new Ethos treatment planning system (TPS). These plans were for patients that clinically required offline plan adaptation due to significant changes in anatomy, which tested the robustness of the ethos oART system to account for such drastic changes. Our optimal approach to plan these patients in Ethos was based on clinical goals/priorities, anatomically derived isodose-shaping helper structures, air density override, additional goals to control hotspots, as well the use of normalization. Online adaptive sessions were simulated using the offline adapted simulation-CTs that were used clinically. Using our developed approach, we found that Ethos generated clinically acceptable initial SEQ plans with comparable planning target volume (PTV) coverage (Dmin,0.03cc = 97.9%, V100% = 98.3%, and D0.03cc = 105.5%) and organs at risk (OAR) sparing. However, Ethos initial SIB plans were inferior to the plans used clinically (for the conventional treatment approach) with lower PTVHigh Dmin,0.03cc = 93.7% and higher hotspot D0.03cc = 110.6%. In Ethos, fixed-field IMRT were higher quality compared to VMAT. The Ethos online adapted plans succeeded in optimizing the dose to the new anatomy in both SIB and SEQ plans. All of this was in a time-efficient manner, ranging from 15 to 30 minutes for the adaptive sessions.

In the fourth chapter, we aimed to investigate the effects of PTV margin decrease for H&N patients that are receiving oART assuming the online adapted plan will reduce uncertainties arising from complex anatomy changes. We used the boost phase of the SEQ plans to generate adapted plans with different PTV margin (5, 3, 2, 1 and 0mm) around the clinical target volume (CTV; generated by a 5mm isotropic expansion around the gross tumor volume or GTV). Online adapted structures drawn by the physician and adapted plans, based upon using a PTV margin of 3mm around the CTV, were available from the simulation of the adaptive sessions in the previous study. Generation of the adapted plans with different PTV margin was possible in the initial planning workspace using the data (images and structures) from the simulated adaptive sessions. A validation of the methodology was investigated to show the similarity of the optimization and dose calculation results in the initial and online adaptive workspace. The difference in dose between the original and the recalculated/reoptimized initial and adapted plans was less than 1%. The adapted plans when using decreased PTV margins demonstrated a consistency in preserving target coverage while decreasing OAR dose by 6%, 10%, 13% and 22% on average when going from 5mm to 3mm, 2mm, 1mm, and 0mm PTV margins, respectively.

In the fifth chapter, based on studies showing that low dose radiotherapy (LDRT) may improve immunotherapy response for polymetastatic non-small cell lung cancer (NSCLC), a workflow/planning strategy to deliver LDRT to all sites of polymetatstic disease using both conventional planning and the AI-based Ethos oART platform is developed and evaluated. We used PET/CT data for ten polymetastatic NSCLC patients to delineate GTVs based on PET standardized-uptake-value (SUV) thresholding. PTVs were created with 1cm margin to account for setup/contour uncertainties and organ motion. Plans were generated in both conventional and the Ethos oART systems based on our developed approaches. For conventional planning, out approach was based on defining the number of isocenters and their distributions by sectioning the region of the patient containing targets evenly in the cranio-caudal direction based on the Halcyon maximum field size of 28cm. One to two arcs were created for each isocenter, and the conventional planning objectives were used to achieve the best coverage and conformality. For the Ethos oART system, PTVs were grouped in RT intents to best encompass all targets without targets overlapping within multiple RT intents. Ethos plan optimization relied on assignment of clinical goals for PTVs. Simulation of online adaptive sessions was done in the Ethos emulator using a subsequent diagnostic scan (when available) or a low-resolution deformation of the initial diagnostic CT (from the PET/CT study) to approximate anatomical changes representative of real clinical scenarios. We found that our developed initial planning approach to treat all sites all polymetastatic disease in both a conventional system and the Ethos oART system succeeded in generating acceptable plans. Plans generated within the Ethos system provided more conformal dose to targets. Similarly, the Ethos online adapted plans succeeded in optimizing the dose to the new disease and maintained conformality in an average total adaptive session time of 26min. The duration of every step of the online adaptive workflow as a function of number of targets showed a low correlation coefficient for influencer generation and editing while this showed a high correlation coefficient for target generation, target editing and plan generation. This timing data also provided an estimate of total adaptive session time as a function of number of targets to guide future clinical implementation of this type of treatment. The feasibility of conventional planning/treatment with Raystation/Halcyon is demonstrated with this work and serves as a benchmark comparison for the developed methodology using Ethos. The efficiency gains when utilizing semi-automated planning/online-adaptive treatment with Ethos for immunostimulatory LDRT conformally delivered to all sites of polymetastatic disease is also illustrated.

In the sixth chapter, we present the major findings of our work with the Ethos oART system for H&N and polymestatic disease in IV NSCLC cancer. We also discuss potential future directions including the evaluation of H&N adapted plan dose when using variable PTV margin on the offline adapted anatomy assuming they represent the ground truth and what these patients would actually receive from these plans including the uncertainty introduced by the online anatomy delineation.

Download

Included in

Physics Commons

Share

COinS