Residency · Residency · Radiation Oncology
Adaptive Radiation Therapy
Overview
Adaptive radiation therapy (ART) involves modifying the treatment plan during the course of radiation to accommodate anatomic changes that occur between the initial simulation and treatment or between individual treatment fractions. Unlike standard radiation therapy, which relies on a single plan created at simulation and delivered unchanged throughout the treatment course, ART recognizes and addresses temporal anatomic variations. These variations include changes in tumor volume, organ position, patient weight, and cavity filling, all of which can degrade the quality of the original plan. ART is implemented primarily through two paradigms: offline ART, where replanning occurs at predetermined intervals, and online ART, where replanning happens at each fraction.
Rationale for Adaptive Therapy
Sources of Anatomic Change
Anatomic changes during radiation therapy arise from several sources. Tumor shrinkage is a common phenomenon, especially in head and neck cancers, where tumors can reduce in size by 1-2% per day, resulting in significant volume changes by the third or fourth week of treatment. Weight loss, frequently observed in head and neck cancer patients, often ranges from 5-10% of body weight and leads to alterations in body contour, shifts in parotid gland position, and changes in mask fit. Organ motion also contributes to anatomic variability; for example, bladder and rectal filling can vary substantially in prostate and cervical cancer patients, while gastric distension and bowel peristalsis affect abdominal organs. Changes in surgical cavities, such as shrinkage of the breast lumpectomy cavity or accumulation and resolution of pleural effusions, further modify anatomy. Radiation-induced tissue edema and inflammation can alter the geometry of targets and organs at risk (OARs). Finally, patient positioning may change progressively due to pain or debilitation, complicating reproducibility.
Dosimetric Consequences of Non-Adaptation
Failing to adapt the treatment plan to these anatomic changes can have significant dosimetric consequences. Tumor shrinkage may cause the high-dose region to extend beyond the reduced target volume, potentially missing tumor tissue that has shifted. Conversely, OARs that move into the high-dose region, such as parotid glands shifting medially as the tumor shrinks, may receive higher doses than initially planned. Overall, the plan optimized on the simulation anatomy may no longer be optimal when applied to the anatomy present on the treatment day, leading to plan degradation.
Offline Adaptive Radiation Therapy
Concept
Offline ART involves replanning at predetermined intervals, such as weekly, or when significant anatomic changes are observed. This process requires acquiring a new CT or cone-beam CT (CBCT), redrawing contours, and generating a new treatment plan that is verified before replacing the original plan for subsequent fractions.
Workflow
The offline ART workflow begins with monitoring daily CBCT images for anatomic changes. When a significant change is detected, a replanning CT scan is acquired. Target volumes and OARs are then re-contoured on this new imaging, followed by creation and optimization of a new treatment plan. After performing quality assurance (QA) on the plan, it is implemented for the remaining treatment fractions. Dose from the original and revised plans is summed to track cumulative dose accurately.
Triggers for Replanning
Replanning is typically triggered by tumor shrinkage exceeding 1-2 cm or a volume reduction greater than 20-30%. Weight loss over 5% or visible changes in body contour that affect mask fit also prompt replanning. Shifts in parotid gland position that result in doses exceeding constraints, significant changes in pleural effusion altering lung volume or position, and consistent setup shifts indicating systematic anatomic changes are additional indications.
Advantages and Limitations
Offline ART allows correction of accumulated dosimetric drift during treatment. However, because planning takes hours to days, changes are only partially addressed, and further anatomic changes may occur before the new plan is delivered. The process is labor-intensive, requiring physician re-contouring, planner re-optimization, and physics QA for each replan. In clinical practice, typically one to three replans are performed per treatment course.
Online Adaptive Radiation Therapy
Concept
Online ART adapts the treatment plan at each fraction based on the patient's anatomy of the day. The entire workflow—including imaging, contouring, optimization, QA, and delivery—occurs while the patient remains on the treatment table. This approach requires rapid imaging modalities such as CBCT or MRI, fast auto-contouring algorithms, rapid plan optimization, and streamlined QA processes.
Workflow (Typical MR-Linac Online ART)
The typical online ART workflow on an MR-linac begins with patient positioning on the treatment table, followed by MRI acquisition, which takes approximately 2 to 5 minutes. Daily anatomy is contoured using AI auto-segmentation with physician review and editing, a process lasting 5 to 15 minutes. The plan is then adapted to the daily anatomy using one of two methods: Adapt-to-Position (ATP), which shifts the isocenter and adjusts beam geometry without full re-optimization and is faster, or Adapt-to-Shape (ATS), which involves full re-optimization based on new contours and is more accurate but slower. Online QA, including independent dose calculation, takes 1 to 2 minutes. After physician approval, treatment is delivered with real-time MRI monitoring. The total on-table time typically ranges from 30 to 60 minutes.
Platforms for Online ART
Several platforms support online ART. The Elekta Unity MR-Linac integrates a 1.5T MRI with a 7 MV flattening filter-free (FFF) linear accelerator and provides real-time MRI cine during delivery. The ViewRay MRIdian combines a 0.35T MRI with a 6 MV FFF linac and offers real-time tumor tracking with gating. The Varian Ethos system uses CBCT-based online ART with AI-driven contouring and rapid plan adaptation but does not include MRI. The Accuray Radixact platform employs megavoltage CT (MVCT)-based adaptation on a helical tomotherapy system.
Clinical Sites Most Benefiting from Online ART
Online ART is particularly beneficial for sites with significant inter-fraction anatomic variability. In pancreatic cancer, the position of the tumor varies considerably between fractions, and proximity to the stomach and duodenum necessitates GI organ-at-risk (OAR)-aware daily plans. Cervical cancer treatment benefits from adaptation due to dramatic uterine position changes with bladder and rectal filling. Prostate cancer patients experience rectal filling changes that alter prostate position and shape. Bladder volume changes substantially between fractions, shifting the target (bladder wall tumor) position. Liver tumors are affected by respiratory motion and progressive volume changes during treatment. Rectal cancer patients exhibit tumor regression and mesorectal anatomy changes during chemoradiation. In head and neck cancers, online ART is less commonly used due to complexity but may be considered in selected cases with rapid tumor regression.
Dose Accumulation and Deformable Registration
Dose Accumulation
When treatment plans change during the course of therapy, it is essential to track the cumulative dose delivered to targets and OARs. This is achieved by mapping the dose from each plan segment onto a reference anatomy using deformable image registration (DIR). The accumulated dose is then calculated as the sum of all mapped dose distributions.
Deformable Image Registration
Deformable image registration employs mathematical algorithms to map voxels between two image sets, accounting for non-rigid anatomic changes. This technique enables contour propagation, allowing auto-contouring on daily images based on reference contours, and facilitates dose mapping from daily anatomy back to the reference anatomy for dose accumulation. However, DIR has accuracy limitations; it can produce physically implausible deformations, and validation remains challenging. Typically, DIR accuracy is around 2-3 mm but can be worse in regions involving sliding motion, such as the lung-chest wall interface.
The Resource and Workflow Debate
Arguments for Broad Adoption
Proponents of broad ART adoption argue that every patient's anatomy changes during treatment, so all patients could theoretically benefit. The technology is rapidly maturing, with AI contouring reducing the physician time burden. Online ART enables dose escalation by reducing margins, potentially improving clinical outcomes. Certain sites, such as pancreas, cervix, and bladder, demonstrate compelling dosimetric advantages with ART.
Arguments for Selective Use
Opponents highlight that online ART significantly increases on-table time, extending treatment sessions from 10-15 minutes for standard volumetric modulated arc therapy (VMAT) to 30-60 minutes. It also demands increased physician, physicist, and therapist time per fraction. The capital costs of MR-linac or other adaptive platforms are substantial. Moreover, there is limited prospective evidence that dosimetric improvements from ART translate into improved clinical outcomes. Not all anatomic sites exhibit sufficient inter-fraction variation to justify the workflow burden, and fixed-anatomy sites such as brain and spine derive minimal benefit.
<image>A side-by-side anatomic comparison showing a head and neck cancer patient at simulation (left) versus week 5 of treatment (right). The simulation CT shows the original GTV, parotid gland positions, and body contour. The week-5 CT shows substantial tumor shrinkage, medial shift of the parotid glands toward the high-dose region, and reduced neck diameter from weight loss. Dose color wash on both images demonstrates that the original plan applied to the week-5 anatomy results in parotid overdosage and suboptimal target coverage. A replanned dose distribution on the week-5 anatomy restores optimal dosimetry.</image>
<image>A step-by-step workflow diagram for online adaptive radiation therapy on an MR-linac. Six sequential panels show: (1) patient setup and initial MRI acquisition, (2) AI auto-segmentation with contours overlaid on the MRI, (3) physician review and editing of contours on the MRI, (4) plan re-optimization with updated dose distribution, (5) independent QA dose calculation check, and (6) treatment delivery with real-time MRI cine monitoring showing the tumor within the treatment field. A clock icon at each step indicates approximate time (2 min, 5 min, 10 min, 5 min, 2 min, 10 min).</image>
Key Clinical Pearls
Not every patient requires adaptive therapy; it is important to identify those most likely to benefit based on the expected magnitude and clinical significance of anatomic changes during their treatment course. Offline replanning, performed one to three times during treatment, represents a pragmatic middle ground for most departments and captures the majority of dosimetric benefit, particularly for head and neck patients experiencing significant weight loss or tumor shrinkage. Online ART is most compelling for abdominal and pelvic sites where inter-fraction organ motion is the dominant source of geometric uncertainty and where dose-limiting gastrointestinal structures lie immediately adjacent to the target. AI auto-segmentation is a key enabler of online ART; without it, physician contouring time would make daily adaptation impractical. However, AI-generated contours must be carefully reviewed, as accepting incorrect auto-contours defeats the purpose of adaptation. Dose accumulation using deformable registration provides an estimate rather than ground truth, so cumulative dose metrics should be interpreted with appropriate caution, especially in regions undergoing large deformation. The future of ART is likely to be risk-stratified, with standardized pathways triaging patients to non-adaptive, offline adaptive, or online adaptive workflows based on anatomy, disease site, and expected benefit.
References
- Yan D et al. "Adaptive radiation therapy." Phys Med Biol. 1997;42(1):123-132.
- Sonke JJ, Belderbos J. "Adaptive radiotherapy for lung cancer." Semin Radiat Oncol. 2010;20(2):94-106.
- Henke LE et al. "Phase I trial of stereotactic MR-guided online adaptive radiation therapy (SMART) for the treatment of oligometastatic or unresectable primary malignancies of the abdomen." Radiother Oncol. 2018;126(3):519-526.
- Glide-Hurst CK et al. "Adaptive radiation therapy (ART) strategies and technical considerations: a state of the ART review from NRG Oncology." Int J Radiat Oncol Biol Phys. 2021;109(4):921-937.
