Continuous Monitoring of Prostate Position During Radiotherapy

Prostate cancer now accounts for one third of all new cancer diagnoses in men and approximately 30% of men will have external beam radiotherapy as their primary local therapy. Prostate motion during radiotherapy can be divided into interfraction and intrafraction motion. Interfraction motion has been well established and has been largely overcome by daily online image verification with either ultrasound, online CT or implanted fiducial markers, however motion during the radiation beam on time (intrafraction motion) is not corrected and can be the cause of significant errors in radiation dose delivery.

Intrafraction motion: Movement of the prostate after initial treatment setup has been termed intrafraction motion. Estimates of the magnitude and frequency of this motion were initially made using continuous Magnetic Resonance (MR) imaging. Padhani et al reported 16% of patients had > 5mm anterior:posterior motion when imaged for 7 minutes with similar results reported in subsequent MR cine studies, Intrafraction motion can be secondary to organ motion such as bladder filling, respiration or moving rectal gas, or can be due to physical patient motion. With the availability of real-time prostate tracking, clinical data is available to quantify the magnitude and frequency of motion. An early report from Kupelian et al using continuous radiotransponder positioning7 described 41% of fractions with >3mm of motion and 15% > 5mm of motion for > 30 seconds. The risk of motion was noted to increase with longer treatment time.

Our own data using offline autosegmentation of the fiducial marker position of 10 patients showed 38% of fractions >1mm, 4.7% of fractions >3mm and 1.7% of fractions >5mm instantaneously during treatment delivery of approximately 2.5 min 8.

Significance of correcting for motion:

If radiation dose is recalculated for each individual fraction and adjusted for intrafraction motion it is possible to estimate the real dose delivered to the target and compare this to the desired dose. This comparison gives a robust model of the potential benefit for real- time tracking and adjustment for the motion. Overview of kilovoltage intrafraction monitoring (KIM): Kilovoltage intrafraction monitoring is a novel real-time tumour localisation modality. It involves a single gantry- mounted kV x-ray imager (which is widely available on most linacs) acquiring 2D projections of implanted fiducial markers. As the treatment gantry rotates around the patient during treatment , the kV imager acquires 2D projections of the prostate . The fiducial markers are segmented using an in-house developed software package. 3D positions are determined via maximum likelihood estimation (MLE) of a 3D probability density.

In previous work (Ng et al, 2012) we have applied the KIM method with offline segmentation to calculate the 3D prostate trajectory after treatment. In the present study we are utilising online marker segmentation to enable live trajectory creation during treatment delivery with a processing time less than 1 second. With the circa real-time trajectories we are able to gate the delivery so to maximise dose to the tumour, or track with Multi-Leaf Collimators (MLC) to follow the prostate motion. The latter involves complex interaction with the beam delivery system and is not part of this study. The treatment can be gated based on a pre-set tolerance. Using our earlier data we modelled several gating criteria 3mm/5s, 3mm/10s and 3mm/15s based on excursions along individual axes and also the radial excursion. Across the 10 patients, a tolerance of 3mm/5s was shown to be efficient, introducing only 24 gating events (from 268 treatments) and safe, with the smallest time for excursion (5s).

The KIM method does introduce extra radiation dose to the patient of approximately 65mSv per treatment localised to the prostate 8. Standard treatment doses for prostate radiotherapy are 80 Gy. The amount of imaging dose will be dependent on the imaging field size, frame rate of acquired images, treatment field size, kV energy used and method of treatment delivery (Volumetric Modulated Arc Therapy [VMAT], Intensity Modulated Radiotherapy [IMRT], Stereotactic Boost Radiotherapy [SBRT]).9 We will minimise the imaging dose from our previous study 8 by:

• reducing the field size to encompass only the fiducials and a small margin per patient so that the imaging dose is delivered inside the treatment volume (expect 20-40% dose saving);
• the beam quality is maximised to 125 kV;
• patient dimensions are limited as the larger patients create more scatter radiation (that decreases image quality) and absorb more radiation. This, combined with treatment field size as below, may allow us to use a lower frame rate for image acquisition, down from 10 fps to 5 fps, which will have a proportional effect on dose;
• patients with nodal areas for treatment are excluded as the treatment fields are larger and this reduces image quality. By excluding these larger patients we may be able to use 5fps rather than 10 fps imaging combined with the patient dimension as above;
• VMAT is shown to require less treatment time than IMRT delivery, and SBRT is shown to require less total time than conventional fractionation (where total treatment time is proportional to imaging dose). We expect to see a dose reduction with the VMAT-SBRT protocol compared to VMAT of 20-40%.

Together with the imaging dose, we need to consider and weight the gains from intrafraction monitoring and gated treatment. These gains include improving the accuracy of delivered dose, so that the planned treatment dose is delivered efficiently to the tumour, and the geometric accuracy that would allow reduction in the safety margin introduced to compensate for treatment. We have demonstrated significant improvements in tumour dose from 60% to 95% of the intended dose distribution by gating with 3mm/5s tolerance as shown in Figure 1.

Verification of KIM clinical dynamic localization accuracy using kV/MV triangulation:

In order to evaluate the dynamic localisation accuracy of the KIM method, 3D positions determined by KIM can be compared to kV/MV triangulation. kV/MV triangulation. Triangulation provides an independent measure of the prostate location.

Failure Mode and Effects Analysis (FMEA) and Quality Assurance (QA) protocol An FMEA has been performed identifying potential failure modes within the additional KIM workflow. The standard workflow has established quality assurance measures in place. All KIM failure modes have been mitigated in numerous layers by quality assurance measures, clear design of user interface, clear designation of staff roles, staff education, and exclusion criteria for patient selection.