On-Line Adaptive Radiation Therapy: Feasibility and Clinical Study

Taoran Li, Xiaofeng Zhu, D Thongphiew, W. Robert Lee, Željko Vujašković, F Yin, Qiuwen WuJournal of Oncology 2010 · DOI

TL;DR

This paper presents a novel on-line Adaptive Radiation Therapy (ART) system for prostate cancer that uses daily cone-beam CT imaging and a fast reoptimization algorithm to adapt treatment plans in real-time. The system demonstrates significant improvements in target coverage and organ-at-risk sparing compared to standard repositioning techniques, while also showing comparable quality to manually generated new plans.

What Problem Does This Paper Address?

External beam radiation therapy (EBRT) for prostate cancer patients faces a significant challenge due to interfractional anatomical variations. Organs like the bladder and rectum can change in shape and position daily, leading to potential under- or over-irradiation of the prostate tumor (CTV) and surrounding healthy tissues. Traditionally, this variability is managed by adding margins to the planning target volume (PTV), which can lead to unnecessary irradiation of healthy organs. While Image-Guided Radiation Therapy (IGRT) has improved patient positioning, it often relies on simple translational corrections and doesn't fully account for complex anatomical changes. This paper addresses the need for a more sophisticated approach that can adapt treatment plans dynamically to these daily anatomical variations, ensuring accurate dose delivery to the target while minimizing dose to organs-at-risk.

Methods

The core of the proposed ART system is a fast intensity-modulated radiotherapy (IMRT) plan reoptimization algorithm. This algorithm takes daily cone-beam CT (CBCT) images, performs deformable registration to compute a 3D deformation field, and then deforms the original dose distribution to match the daily anatomy. This "goal daily dose map" is then used by a linear programming model to reoptimize the fluence maps, generating new intensity maps for the treatment beams. This process is designed to be fast, completing within two minutes.

The study also introduces a plan quality evaluation tool that uses machine learning to predict achievable dose-volume histograms (DVHs) based on a database of previous treatment plans and patient anatomy. This tool helps assess the quality of the adapted plans by comparing them to predicted ideal DVHs.

Finally, a dual-technique clinical implementation scheme is proposed. This scheme prioritizes using existing plans from a patient-specific database for fractions where anatomy is similar to previous plans. If no suitable existing plan is found, the system resorts to the on-line reoptimization technique. This aims to minimize the frequency of computationally intensive reoptimization while maintaining high treatment quality.

Datasets

The study utilizes data from 6 prostate cancer patients, with each patient contributing 1 planning CT and 5 daily CBCT image sets, totaling 30 treatment fractions for the comparison of adapted vs. repositioned plans. For the plan quality evaluation, a database of 198 existing treatment plans was used to train the machine learning model. For the clinical implementation evaluation, data from 18 patients were simulated.

Key Insights and Ideas

  • Real-time adaptation is feasible and beneficial: The developed ART system can adapt treatment plans quickly enough for clinical use, demonstrating significant dosimetric advantages over standard methods.
  • Deformable registration is crucial: Accurately capturing daily anatomical changes, not just rigid shifts, is key to effective adaptation.
  • A "dual-technique" approach balances efficiency and quality: By intelligently selecting between existing plans and on-line reoptimization, the system can reduce computational burden without compromising treatment outcomes.
  • Predictive DVH modeling aids quality assurance: The machine learning-based DVH prediction tool offers an objective way to assess plan quality and identify areas for improvement.

Key Results

  • Improved CTV Coverage: Adapted plans (Adapt) consistently achieved full CTV coverage (D99 within ±2.5% of prescription dose), whereas repositioned plans (Soft) showed significant fluctuations (45%-103%) and compromised coverage in 46% of fractions.
  • Enhanced OAR Sparing: Statistical analysis showed that adapted plans generally provided better sparing of the bladder and rectum compared to repositioned plans, particularly in low- and high-dose ranges.
  • Comparable to New Plans: Head-to-head comparisons between adapted plans (Adapt) and manually generated new plans (New) revealed highly comparable CTV coverage (D99 deviations within -0.3% to 1.4%) and similar control of low- and high-dose regions.
  • Efficiency Gains: The dual-technique scheme demonstrated a reduction in the need for reoptimization as more fractions were delivered, with patients increasingly treated using existing plans from their database.

Limitations and Future Work

  • Contouring Consistency: The study relied on retrospective contouring by a single experienced physician to ensure consistency between planning CT and daily CBCT images. This highlights the importance of robust contouring protocols in clinical ART implementation.
  • Image Quality: CBCT image quality, particularly soft-tissue contrast, can be a challenge for accurate contouring.
  • Integration Challenges: Integrating the developed system with existing clinical treatment planning and delivery platforms (e.g., Eclipse and ARIA) requires configuring data interfaces.
  • QA Efficiency: Developing efficient quality assurance (QA) techniques for ART plans is crucial for maximizing efficiency.
  • Radiobiological Indices: Future work should extend the evaluation to radiobiological indices like EUD, TCP, and NTCP to fully assess the clinical benefit.
  • Patient Instructions: The paper notes that patient instructions and preparation (e.g., full bladder) can influence anatomical variations, and further validation against such attempts is needed.

Summary updated

Abstract

The purpose of this paper is to evaluate the feasibility and clinical dosimetric benefit of an on-line, that is, with the patient in the treatment position, Adaptive Radiation Therapy (ART) system for prostate cancer treatment based on daily cone-beam CT imaging and fast volumetric reoptimization of treatment plans. A fast intensity-modulated radiotherapy (IMRT) plan reoptimization algorithm is implemented and evaluated with clinical cases. The quality of these adapted plans is compared to the corresponding new plans generated by an experienced planner using a commercial treatment planning system and also evaluated by an in-house developed tool estimating achievable dose-volume histograms (DVHs) based on a database of existing treatment plans. In addition, a clinical implementation scheme for ART is designed and evaluated using clinical cases for its dosimetric qualities and efficiency.
BibTeX
@article{li2010line,
  author = {Taoran Li and Xiaofeng Zhu and D Thongphiew and W. Robert Lee and Željko Vujašković and F Yin and Qiuwen Wu and Fang-Fang Yin and Q. Jackie Wu and Q. Jackie Wu},
  title = {{On-Line Adaptive Radiation Therapy: Feasibility and Clinical Study}},
  journal = {Journal of Oncology},
  year = {2010},
  doi = {10.1155/2010/407236},
}
Figure