DGRT: it's the dose that matters

Various techniques are under investigation for image-guided radiation therapy (IGRT), including kilovoltage cone-beam CT (kV CBCT), megavoltage cone-beam CT (MV CBCT) and, further ahead, MRI and ultrasound-based methods. These modalities can all monitor any deviations from the expected patient position and anatomy. But according to Jean Pouliot of the University of California, San Francisco (UCSF), what's really important is how such geometric and anatomical changes impact upon the delivered dose.

With this in mind, Pouliot's research group is investigating the concept of dose-guided radiation therapy (DGRT), in which MV CBCT images are used to reconstruct the delivered dose, and dosimetric considerations comprise the basis for treatment modification. Tami Freeman spoke to Jean Pouliot, professor of Radiation Oncology and director of UCSF's Division of Medical Physics, to find out more.

TF: What is the main driver for implementing image guidance during radiotherapy?

JP: The need for dose escalation to the tumour, the reduction of the number of fractions used to deliver the dose, and anatomical changes during the course of the treatments are all sources of motivation to introduce IGRT.

However, it is the ability to accurately deliver highly conformal dose to the clinical target, provided by the widespread use of intensity-modulated radiation therapy, that makes it imperative to verify the patient set-up and implement the use of IGRT routinely. The early adopters of IGRT clearly demonstrated the difference between intended and delivered dose.

How is MV CBCT currently employed during patient set-up?

Once the patient is positioned on the treatment couch, an image is obtained with the MV CBCT by rotating the gantry around the patient. Just like a regular CT, the imaging system acquires a large number of projection images and reconstructs a tomodensitometric image (CT image) that provides a 3D view of the patient in the treatment position. This image is then automatically or manually compared with a reference image (usually the CT that was used for dose planning) and their registration is used to determine whether a patient set-up correction is required. The set-up is then adjusted and the dose can be delivered accurately. This IGRT process – image acquisition, registration and set-up correction – is performed rapidly in 2–4 minutes.

The landmarks and process used to register the reference image with the MV CBCT are specific to the anatomical site treated. For instance, for head-and-neck treatment, bony structures such as the cervical vertebrae provide a reproducible landmark. Positioning of the prostate requires implantation of radio-opaque gold markers. The prostate changes position within a few minutes and requires IGRT daily.

How can MV CBCT images be used to reconstruct the delivered dose?

IGRT allows the precise positioning of the patient. As a by-product, it provides a 3D image of the patient in the treatment position. We realized very early during the development of MV CBCT that these images could be processed to allow precise dose calculation. This opens up the possibility for in situ dose calculation, i.e., evaluating the delivered dose for the actual position and shape of the anatomy just a few seconds before dose delivery. We no longer need to assume that the intended dose is the delivered dose. We can verify this, and act accordingly when important differences are observed.

How is this information used to adapt the remaining treatment fractions?

IGRT allows imaging of anatomical changes during radiation therapy. When a tumour regresses or a patient loses weight, one can wonder if the initial dose plan is still valid. But the impact of these changes is also dependent on many other factors, including the dose distribution, the beam arrangement, etc. So the real question is not whether the anatomy is changing, but whether this matters? In other words, what is the dosimetric impact of these observed changes?

Dose recalculation with MV CBCT, or dose-guided radiation therapy (DGRT), can accurately answer this question. If the dose tolerance to an organ-at-risk is compromised, then one can generate a new plan to accommodate the changes. Several clinical studies will be required before a systematic approach is developed. But at least now we have the means to answer these crucial questions. Again, instead of asking if the patient is correctly positioned, one can now ask if the IGRT protocol ensures target dose coverage. Studies are currently being conducted with this line of questions.

What would you say is the main benefit of dose-guided radiation therapy?

DGRT brings the focus back to the main objective of radiation therapy, i.e., to deliver dose to the tumour. In other words, the end-point is not to position the patient, align the beams, etc. Although all of those steps are necessary, the real end-point is to ensure that the intended dose is delivered to the correct location. And we can validate this by evaluating the delivered dose every time the patient is treated. In addition to providing a better level of quality assurance, this provides a record of what was really delivered to the patient. This information may prove extremely useful if any questions arise, or if the patient needs to be retreated months or years later.

Is DGRT being used clinically yet? If not, what processes are required to bring this technique into the clinic?

I consider DGRT as an extension of IGRT where dose considerations (not image or geometry) are used to decide on the next action. There are many different scenarios of DGRT. In its simplest form, the images used for IGRT could be processed off-line and the delivered dose calculated and archived. These dose distributions could be reviewed for quality purpose. In our clinic, we have the ability to do this. When a doubt is raised as to the extent of an anatomical change, weight loss, tumour regression and so on, then the delivered dose is calculated and compared with the planned dose.

With current systems, many steps are involved before one can display the delivered dose distribution. These include, to name a few: image correction for cupping effect, electron density calibration, image registration, copy of the initial beam configurations to the MV CBCT image, dose calculation, copy and adjustment of the organ definition (contours) from the planning CT to the MV CBCT, computation of dose-volume histogram, dose difference and other values useful for dose evaluation.

The proofs of concepts have been established. We have built application tools to automate some of these steps and facilitate the workflow process. The implementation of DGRT in the clinic requires that those steps are integrated with the commercial platforms. Some vendors are making good progress along this line.

How do you see DGRT being developed and implemented in the future?

Safety in radiation therapy is in everyone's mind these days. DGRT enables the evaluation and recording of the in situ delivered dose. I think it will eventually become mandatory to record the delivered dose for each patient. This will enable a better correlation between dose response and delivered dose (instead of intended dose).

A small fraction of the dose could be delivered and verified with the DGRT process before the complete dose is delivered. This could be particularly useful when a high dose is delivered in a single or a small number of fractions.

Access to information about the dose-of-the-day will open the doors to many areas of research and clinical improvements in treatment verification and dose accumulation. DGRT can provide valuable feedback about the dose delivered during each treatment fraction. Finally, trends in dose differences can reveal important consequences of anatomical changes or set-up errors.