An adaptive strategy for radiation therapy in the pelvis: plan selection
Rianne de Jong, from The Netherlands, describes how radiation therapists have established a greater role in plan selection
Over the last decade radiation therapy has evolved rapidly. With state-of-the-art treatment planning techniques like IMRT (Intensity Modulated Radiation Therapy) and VMAT (Volumetric Modulated Arc Therapy) we are able to sculpt the dose tightly around the target volume, thereby reducing the irradiation dose in healthy tissue.
To make an irradiation plan, a planning CT scan is acquired prior to treatment, on which the target volume is defined. The same irradiation plan is used during the entire course of treatment, but the shape and position of the target volume within the patient typically varies slightly from day to day. To ensure that the entire tumor is irradiated every treatment day, safety margins are applied around the target volume. Besides using safety margins, image guided radiotherapy (IGRT) can be applied to verify and correct the position of the target volume. An IGRT technique, which is currently widely available, is the use of cone beam CT (CBCT) scans, which allows soft tissue visualisation.
Special challenge of the pelvic area
These 3D (or 4D) CBCT scans are acquired daily just prior to treatment at the treatment machine, and are compared to the reference image, e.g. the planning CT. If the target volume is rigid, the positional error, found by registering the two images, can be corrected with a treatment couch adjustment. This in turn allows for minimal safety margins to maximally spare the healthy surrounding tissues. However, the pelvic area offers a special challenge. The target volumes in the pelvic area do not so much move rigidly from day to day, but are prone to large deformations. These deformations are mostly caused by different bladder and bowel fillings and cannot be corrected with a treatment couch adjustment. Incorporating these deformations into the safety margin leads to very large margins.
Reducing the margins in radiotherapy of the pelvic area therefore requires management of deformations. Prevention of deformations with the use of a drinking protocol, to manage bladder filling, and dietary instructions, to manage bowel motion, have been shown to be of only limited success. Currently, when deformations are seen on the CBCT scan, patients can be sent back to the waiting room for some time, e.g. to further increase bladder filling. However, this is time consuming, a logistic challenge and, in many cases, still does not provide the desired result.
A new approach – plan selection
Therefore a new approach to cope with these deformations is developed: plan selection. This is an adaptive strategy where multiple irradiation plans are made prior to treatment. The plans are tailored to a range of possible deformations of the target volume of the individual patient. Each individual plan has a smaller safety margin compared to the margin necessary when only a single plan is used, and with that, the dose to the healthy surrounding tissues will be decreased. A plan selection strategy is sometimes also referred to as ‘library of plans’, or ‘plan of the day’.
With this strategy, the CBCT scan, acquired daily prior to treatment, is used to select the plan which is most appropriate for the geometry of the target volume for that specific fraction. This is a particularly effective strategy for bladder and cervical cancer, as the deformation of the target volumes can mostly be predicted by the volume of the bladder. To create multiple plans, often a full and an empty bladder pretreatment CT scan are acquired, from which a patient-specific model for the bladder filling is derived. This model is used to create intermediate shapes for the target volume. Typically, between two and five plans are made, to limit not only the workload for treatment planning but also the complexity of selecting a plan at the treatment machine.
In addition to bladder and cervical cancer, large deformations are also found in the target volume for rectal cancer. However, for these patients target volume deformation is not so much driven by the difference in bladder filling but mostly by changes in the rectum volume. Because of this, creating multiple plans based on varying the bladder filling is not very useful. Creating multiple plans based on empty and full rectum CT scan is considered challenging and “patient unfriendly”. However, plan selection can still be applied but a slight change in approach is necessary. A single planning CT scan can be used, on which multiple safety margins of different sizes around the deforming part of the target volume are applied. This results in multiple treatment volumes for which multiple plans can be created.
Obstacles to this strategy
Due to challenges regarding workload and workflow, plan selection is not yet an often-used strategy. First, the algorithms used to create the patient-specific motion models are not commercially available and are difficult to incorporate in existing software and workflows. Also, creating more than one plan increases the workload and is therefore more expensive, but has not yet been proven to be superior in terms of better tumor control or less side effects. Finally, there are two issues with respect to selecting a plan on a daily bases at the treatment machine: Firstly, is the image quality of a CBCT sufficient to select a plan, as the image quality can be hampered by artefacts caused by moving air in the bowel? Secondly, it has to be determined who is responsible for selecting the appropriate plan at the treatment machine. In the Netherlands, a treatment is typically delivered by radiation therapists (RTTs), which includes decision-making within the IGRT procedures. All protocols support this workflow.
Determining a ‘gold standard’
We decided to address these two issues, image quality and responsibility, at the same time and designed an inter-observer study. This observer study (bladder, cervix and rectum separately) started with a lecture by an expert radiation oncologist on target definition followed by a first measurement (baseline). The baseline measurement entailed plan selection by all observers individually, without the possibility of discussion between the observers. All disciplines contributed but RTTs were the majority of observers. After the baseline measurement, all observers and the department’s expert radiation oncologists together determined the ‘gold standard’ during a consensus meeting. During this meeting, all CBCT scans of all patients were discussed, and for each CBCT scan the consensus for the best fitting plan was determined, ultimately based on the decision of the expert radiation oncologists.
One month later, a second measurement was conducted, in which the same observers repeated plan selection for all CBCT scans on the same data set as the baseline measurement. Observers, again, did not have the possibility to discuss with each other, and were blinded to both the gold standard as well as their own selections from the baseline measurement. This process provided both the RTTs and the radiation oncologists with the confidence that the RTTs had gained the expertise to select the appropriate plan based on CBCT scans at the treatment machine. As an example, in the observer study for rectal cancer, accordance between plan selection by the observers and the gold standard was 69% in the baseline measurement, and 75% in the second measurement. The RTTs had similar accordance with the gold standard as the physicians and the physicists.
RTTs take responsibility
We therefore decided that responsibility for plan selection was with the RTTs, and the observer studies had served as a training. During the clinical implementation process, for the first 20 patients a radiation oncologist, a clinical physicist and a trained RTT were present during plan selection in the first week of treatment. From week two onwards, only two RTTs, of which one was a trained RTT, were responsible for plan selection. This process was used in our clinic to implement plan selection for bladder, cervix and rectum treatments.
With this strategy of plan selection, which entails having multiple plans available at the treatment machine to accommodate deforming target volumes, we have successfully reduced safety margins for bladder, cervix and rectum treatments. Whether the resulting dose reduction to the surrounding healthy tissue will result in fewer side effects is a topic for further investigation.
M.A.J. de Jong, Radiation Therapist, Academic Medical Centre, Department of Radiation Oncology, Research and Development IGRT&ART, Amsterdam