Online adaptive planning for prostate cancer radiotherapy is necessary and ready now
Year: 2014
Type: article
Abstract: In linear-accelerator based radiotherapy for prostate cancer, both target and normal tissues are known to change position and shape considerably during a course of therapy. The need to track and compensate for these motions is well established and adaptive planning is widely accepted. Some claim that this should be accomplished online while the patient is lying on the couch waiting for treatment and this is the premise debated in this month's Point/Counterpoint. Arguing for the Proposition is X. Allen Li, Ph.D. Dr. Li is Professor and Chief of Medical Physics in the Department of Radiation Oncology, Medical College of Wisconsin. He is certified in Radiation Oncology Physics by both the Canadian College of Physicists in Medicine and the American Board of Medical Physics. Dr. Li has served for the AAPM in the capacity of chair or member of various subcommittees and task groups including the Biological Effects Subcommittee, TG-74, TG-106, TG-166, and the Editorial Board of Medical Physics, and is a Fellow of the AAPM. He has been a peer reviewer for 14 scientific journals and seven public and private research funding agencies. He has edited a book entitled “Adaptive Radiation Therapy” and has authored over 125 peer-reviewed papers. Dr. Li's research interests range from adaptive radiation therapy, outcome modeling, and recently to MRI guided radiation treatment planning and delivery. Arguing against the Proposition is Qiuwen Wu, Ph.D. Dr. Wu obtained his Ph.D. in Physics from Columbia University, New York, and subsequently worked in the Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, New York, NY, the Department of Radiation Oncology, Virginia Commonwealth Univer-sity, Richmond, VA, William Beaumont Hospital, Royal Oak, MI, and Wayne State University, Detroit, MI, before moving to Duke University Medical Center where he is currently Professor in the Department of Radiation Oncology. He is certified in Therapeutic Radiological Physics by the American Board of Radiology, is a Fellow of the AAPM and is a member of the Board of Editors of the Journal of Applied Clinical Medical Physics. Dr. Wu's major research interests include adaptive radiation therapy, online image guidance, and applications of flattening filter free linacs, for which he has published about 70 peer-reviewed papers. It has been widely reported that interfraction variations in both targets and organs at risk (OARs) can be substantial during radiation therapy (RT) for prostate cancer.1 These variations, including translational and rotation shifts, deformation (e.g., volume and shape changes) and independent organ motions, can be systematic and random in nature. To account for these variations, large CTV-to-PTV margins are necessary to ensure adequate target coverage. These large margins inevitably result in increased doses to the OARs, leading to increased treatment-related toxicities that may prevent safe delivery of more effective and/or socially/economically favorable treatments such as dose escalated RT, hypofractionated RT, or stereotactic body RT (SBRT). To address the negative impacts listed above, a large amount of effort has been expended in recent decades on the development of technologies and strategies to correct for interfraction variations.2 Image-guided RT, where images acquired immediately prior to a treatment are used to guide patient repositioning, is currently the standard practice used to correct translational shifts and rotational errors, if the machine is properly equipped, but does not correct for anatomy deformations and independent organ motions. Adaptive RT (ART) that may be performed online or offline has been introduced to correct for this problem.3 While offline ART may be used to account for systematic variations,4 online ART that generates and delivers a new dosimetric plan optimized based on the anatomy of the day (fraction) can fully account for interfraction variations including both systematic and random (unpredictable) deformations and independent organ motions.5,6 Online replanning needs to be fast so that it can be completed within a few minutes while the patient is lying on the table waiting for treatment. Although such fast planning is generally challenging using conventional planning technologies, adaptive replanning does not need to start completely from scratch. For example, it can start with an initial plan fully optimized from the planning images for the same patient. Technologies on the quality of inroom imaging, image registration and segmentation, plan optimization algorithm, and computing hardware are advancing significantly and rapidly. For example, improved quality of onboard cone-beam CT, addition or integration of diagnostic-quality CT or MRI in the treatment room, graphic-processing unit accelerated autosegmentation and dose calculation,7 rapid plan modification with aperture morphing algorithms,6 and plan adaptation based on previous knowledge or a previously created plan library, are among the technological advances that can speed up adaptive planning significantly. Commercial treatment planning systems including some of these advances are becoming available. Extensive literature indicates that the interfraction deformations of prostate, bladder, and rectum and independent motions between prostate, pelvic nodes, rectum, and bladder during RT for prostate cancer, are generally unpredictable and can be substantial. For example, based on the MRIs acquired during the course of prostate RT, Nichol et al.8 reported that prostate volume could decrease by 20% and its shape could deform by 13 mm. Patients with a transurethral resection of prostate were prone to prostate deformation.8 It is necessary and possible to fully account for these variations by using online adaptive replanning for a treatment strategy requiring a small CTV-to-PTV margin. With online ART, a margin can reach as low as 3 mm, depending mainly on intrafraction variations. Such a small margin would be highly desirable to reduce treatment-related toxicities. In conclusion, online adaptive planning for prostate cancer radiotherapy is necessary and ready now, particularly for cases with large deformations or for dose escalated RT, hypofractionated RT, or SBRT. With advanced onboard imaging systems, standard online image guidance (IG) techniques currently used in many clinics is efficient and can correct setup errors and rigid organ motions. The residues, including uncorrected rotations, execution errors, deformations, and intrafraction motions, are handled by the margins in treatment planning, which are significantly reduced from those used for conventional treatment planning without IG.9 In principle, online planning, either adapted geometrically to the target or dosimetrically to include dose from previous fractions, can compensate additionally for interfractional deformations of both targets and critical organs in real-time while the patient is still on the table, potentially allowing the use of minimal margins.10 However, there are steep challenges ahead, both technically and practically. The crucial process in treatment planning is the delineation of regions of interest (ROIs). Unfortunately, current state-of-the-art online CBCT images are not adequate for this task for prostate cancer treatments.11 Radiation oncologists need significantly longer time and yet with less confidence to manually contour on CBCT than conventional CT. Also, the accuracy of computerized segmentation and deformable image registration algorithms depends heavily on image quality and, therefore, margins for the uncertainty of ROI definition must be added. In addition, online adaptive planning will undoubtedly increase the time between imaging and treatment and therefore needs increased margin for intrafractional motion and deformation, which increase with time.12 The potential savings in margins for online planning can quickly diminish. The prolonged treatment session is not economical either. Many practical factors also limit the use of online planning. The physician is required to approve the ROIs on online CT as well as the treatment plan, which the planner needs to generate quickly. In addition, the physicist needs to validate the plan and perform quality assurance creatively with the patient on the table. In the meantime, the therapists need to check all new plan parameters. Ideally, all these tasks need to be done at the treatment console for efficient communication and minimization of errors. However, this is unrealistic for scheduling in many clinics since it requires additional personnel at the linac at the same time for each prostate patient. There are competing alternatives to handle nonrigid organ motions such as the hybrid strategy combining online IG and offline adaptive planning.13 The advantage is that time-consuming treatment planning is handled offline, the same way as traditional treatment planning. With no added burden to the online process, the online IG is the same efficient process that patients and therapists are used to. Many studies have shown that organ motion and deformation in prostate cancer are patient-specific in nature.14,15 Offline or hybrid adaptive radiotherapy can take advantage of this by analyzing repeated images during early treatments and incorporating them into the treatment planning. Dose guided adaptive radiotherapy and equivalent margin reductions can be achieved.16 In summary, online adaptive planning faces many technical and practical challenges and is not yet ready for clinical implementation. Increased margins for target definition uncertainty and intrafraction motion offset the savings in margin for deformation. In the meantime, clinically proven alternative offline and hybrid adaptive strategies have been shown to be equally as effective. The offline and hybrid adaptive strategies cannot adequately account for random (unpredictable) organ deformations and independent organ motions and hence they are certainly not as effective as the online adaptive approach. They would require larger margins for cases with moderate and large deformations. Also, I have to disagree with Dr. Wu on his point that the additional time for online adaptive planning required in order to increase margins to account for the increased intrafraction variations may diminish the benefits of online replanning. For cases with moderate and large deformations, the interfraction variation is dominant. It has been documented that the dosimetric loss due to the intrafraction changes during the application of online replanning is relatively small compared to the gain from the online replanning. Furthermore, the time required for online replanning can be reduced substantially (to a few minutes) with the recent advances in online imaging, image registration, replanning algorithms, and/or computing technology. Undoubtedly, large efforts and resources are needed for online replanning, similar to any other emerging techniques. However, the benefit of the reduced margins from online replanning is clear. These reduced margins would lead to reduced radiation injury to normal tissues and/or offer opportunities for more effective treatments (e.g., dose escalation, hypofractionation). For prostate radiotherapy, the increased effort can be socioeconomically justified in cases with large deformations or when using hypofractionated RT or SBRT. Both Dr. Li and I agree that there are a variety of substantial interfractional organ motions in prostate cancer patients. We disagree on the effectiveness and efficiency of online adaptive planning to handle them as it stands now. Its advantage over alternatives is too little, and it is too soon for clinical implementation. Many pelvic organ motions, including deformations and intrafraction motions, are not totally random but rather patient-specific; they can be characterized based on a few measurements from previous fractions, and compensated through either offline or hybrid adaptive planning. In principle, in-room MRI could improve the accuracy and shorten the time of ROI delineation. However, either MRI-only based treatment planning must be clinically proven, or CT-MRI registration be shown to be equally as accurate as CT-CT registration. Pretreatment verification of “a new dosimetric plan optimized based on the anatomy of the day” is a practical concern. With the standard of care of prostate radiotherapy moving toward IMRT and VMAT, the online adaptive plan must be verified before treatment as this is required by regulations. This poses considerable challenges while the patient is on the treatment table. Current workflow in many radiation oncology clinics is sequential in nature, in which each task is performed by a specially trained staff member and completion of a task triggers the next one in line. It is not a trivial task to bring physicians, planners, physicists, and therapists together to the treatment console at the same time for each fraction. All this being said, I believe that online planning represents the holy grail of prostate cancer IGRT, and I am all for the research toward its successful application. Pointing out its current deficiencies should provide us better directions to work on in the future.
Source: Medical Physics
DOI: 10.1118/1.4883875
Authors: X. Allen Li, Qiuwen Wu, Colin G. Orton
Institutions: Medical College of Wisconsin, Duke Medical Center
Language: English
Subfield: Radiation
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,360
OpenAlex ID: W2068256080