Normal tissue sparing in a phase II trial on daily adaptive plan selection in radiotherapy for urinary bladder cancer

Abstract

Background. Patients with urinary bladder cancer often display large changes in the shape and size of their bladder target during a course of radiotherapy (RT), making adaptive RT (ART) appealing for this tumour site. We are conducting a clinical phase II trial of daily plan selection-based ART for bladder cancer and here report dose-volume data from the first 20 patients treated in the trial.

Material and methods. All patients received 60 Gy in 30 fractions to the bladder; in 13 of the patients the pelvic lymph nodes were simultaneously treated to 48 Gy. Daily patient set-up was by use of cone beam computed tomography (CBCT) guidance. The first 5 fractions were delivered with large, population-based (non-adaptive) margins. The bladder contours from the CBCTs acquired in the first 4 fractions were used to create a patient-specific library of three plans, corresponding to a small, medium and large size bladder. From fraction 6, daily online plan selection was performed, where the smallest plan covering the bladder was selected prior to each treatment delivery. A total of 600 treatment fractions in the 20 patients were evaluated.

Results. Small, medium and large size plans were used almost equally often, with an average of 10, 9 and 11 fractions, respectively. The median volume ratio of the course-averaged PTV (PTV-ART) relative to the non-adaptive PTV was 0.70 (range: 0.46–0.89). A linear regression analysis showed a 183 cm³ (CI 143–223 cm³) reduction in PTV-ART compared to the non-adaptive PTV (R² = 0.94).

Conclusion. Daily adaptive plan selection in RT of bladder cancer results in a considerable normal tissue sparing, of a magnitude that we expect will translate into a clinically significant reduction of the treatment-related morbidity.

Patients with urinary bladder cancer referred for radiotherapy (RT) in Denmark are primarily elderly patients with severe co-morbidities. Cystectomy is considered the preferred treatment of bladder cancer, however, a recent randomised trial showed improved local control rates in the RT plus concomitant chemotherapy arm, comparable with rates obtained in cystectomy series [1].

Large changes in bladder shape, size and position are occurring frequently during a course of RT, and this has so far been accounted for using large population-based target volume margins in order to ensure target coverage [2]. This approach leads to irradiation of large volumes of the surrounding normal tissues, causing a high risk of in particular gastro-intestinal (GI) morbidity [3,4]. During the last years, a number of advances in RT planning and delivery have been developed and are now introduced clinically, including in particular image-guided RT (IGRT) as well as adaptive RT (ART) [5–18] . The introduction of cone beam computed tomography (CBCT) for daily setup prior to RT revealed a need for an individualised approach to manage the large internal organ motion in bladder cancer patients [19]. These changes are primarily random [20] and treatment planning studies indicate that ART using plan selection from a predefined library may provide considerable normal tissue sparing [9,10,21,22]. In an early simulation study from our group, we suggested a strategy using delineations of the bladder on the CBCTs from the first few treatment fractions to generate a library of treatment plans for each patient [10]. By this approach, patient-specific variation in the volume and shape of the bladder volume was taken into account, to obtain maximal sparing of normal tissues. This strategy was subsequently investigated on repeat CBCTs in patients treated by conventional radiotherapy [9] and was recently taken into a clinical phase II ART trial. In this paper we report on the normal tissue sparing achieved for the first 20 patients included in the trial.

Material and methods

This study includes the first 20 patients treated in the bladder phase II ART trial, during the period October 2012 to November 2013. All patients had muscle invading disease, and one case had tumour extension into the prostate (T4a). The patients had a mean age of 77 years (ranging from 56 to 87 years). Thirteen patients received pelvic lymph node irradiation with a simultaneous boost (SIB) to the bladder; the remaining seven patients received RT to the bladder only, because of age and/or severe co-morbidities. The trial was approved by the Research Ethcs Committee of the Central Denmark Region.

Imaging, CTV and normal tissue definitions

A treatment planning CT scan (planCT) was acquired, with a reconstruction slice thickness of 3 mm, approximately one week before treatment start. All patients emptied their bladder before the CT scan and before each of the treatment fractions. The patients did not receive restrictions on diet and fluid intake.

Throughout the course of therapy (including the daily plan selection ART), the clinical target volume (CTV) was defined as the bladder and the tumour when visible. The rectal volume was delineated from the recto-sigmoid transition or the sacro-iliac joint to the anal canal (inclusive) while the bowel cavity was delineated from the superior border of the fifth lumbar vertebra and inferior to the last slice with bowel segment.

All patients had pre-treatment CBCTs acquired throughout the course of RT for daily setup on bony anatomy. In the first week post-treatment CBCTs were acquired for all patients and for the first 10 patients a weekly post-treatment CBCT was acquired as well.

Planning for first 5 fractions

For the plan used in the first 5 fractions, a population-based margin of 2 cm superior/anterior, 1.5 cm posterior and 1 cm in the other directions was added to the bladder CTV to create the internal target volume (ITV) [2]. In patients receiving loco-regional SIB the ITV_ln was formed by adding a 7 mm margin around the common iliac vessels below the pelvic inlet, the internal iliac vessels to the obturatory canal as well as the external iliac vessels to the acetabula ceiling. Set-up margins of 8 mm superior/inferior and 5 mm in the other directions were added to generate the PTV_large for the bladder and PTV_ln for the lymph nodes.

Initially, volumetric modulated arc (VMAT) plans with one or two 360˚ arcs were optimised for each patient to deliver 60 Gy in 30 fractions to PTV_large and simultaneously 48 Gy to PTV_ln, when lymph nodes were included (Eclipse v.11, Varian Medical Systems, Palo Alto, CA, USA). This plan corresponded to the non-adaptive RT (non-ART) delivered to patients outside the ART trial in the two involved RT departments. The plan aimed to cover 99% of the PTVs (PTV_large and PTV_ln) with more than 95% of the corresponding prescribed doses and to keep the rectal V40Gy below 50% and the bowel V35Gy below 40%. The maximal dose was limited to 107%. All RT fractions were delivered on TrueBeam accelerators (Varian Medical Systems) using CBCT-based set-up on bony anatomy.

Adaptive treatment planning and delivery

Wright et al. suggested several methods for constructing ART volumes based on the bladder contours on the planning CT and a variable number of repeat image sets [10]. In the present clinical trial the target volume for small bladder size was based on the volume contained in at least two of five CTVs as delineated on the plan-CT and the CBCTs acquired during the first 4 fractions, whereas the medium-sized target was based on the union of all five CTVs [9]. The method was originally developed using soft tissue registration on the bladder, but as registration in this study used bony anatomy; an additional margin of 3 mm was therefore added when creating the small- and medium-sized plan selection volumes (PSV_small and PSV_medium).

An isotropic margin of 5 mm was applied to account for intra-fractional changes in bladder shape, size and position for creation of the corresponding PTVs. The two additional adaptive treatment plans were optimised to deliver the same total dose to the bladder (and pelvic lymph nodes, when included) and to comply with the same constraints as the non-ART treatment plan used in the first 5 fractions.

In 14 patients, the three treatment plans met all constraints. In three patients, it was not possible to meet the constraint for the rectum. For the patient with disease extending to the prostate we included the prostate in the CTV. A second patient had unilateral hip replacement that did not allow the constraints to be met and a third patient had very large and flat rectum. In two patients, the bowel cavity constraint was not met because of bowel segments between the bladder and the rectum and thereby the bowel cavity extended to this region. Another patient had a minor deviation from the bowel cavity constraint.

Daily plan selection-based ART was used in fraction 6–30. Initially, a CBCT was acquired and a bony match was performed with visual inspection of the bladder prior to delivery. First, the PSV_small overlaid on the CBCT was evaluated and if the bladder was covered the patient was treated with the small-sized treatment plan. Otherwise, the PSV_medium was evaluated and if the bladder was covered by this volume, the medium-sized plan was selected. If neither of these two PSVs covered the bladder, the large-sized treatment plan was chosen. In very few occasions, the bladder was not covered by the PTV_large and the patient was asked to empty the bladder before a new CBCT acquisition was performed.

All involved radiation therapists (RTTs) went through an e-learning course of four hours as well as a one and a half day hands-on training in plan selection using retrospective data in a virtual reality learning centre [23]. For the first 10 patients, a weekly offline check of the plan selections was performed.

Evaluation and statistical analysis

The individual plan selection procedures were recorded and the ratio of course-averaged PTV (i.e. the PTV volume averaged over each treatment fraction in the ART course, PTV_course) for the adaptive treatment versus the non-ART treatment was calculated for each case. This volume only describes the averaged volume of high dose and not the lymph node target. Furthermore, the total delivered dose was assessed using the planCT and plan summation based on the number of fractions the different plans were used. Assuming a spherical bladder it can be shown that the PTV will increase linearly with CTV, when the radius is large compared to the margin applied. Finally, the change in course-averaged PTV as a function of CTV was analysed using linear regression with ART as a binary covariate.

As the pelvic organs are very mobile there is considerable uncertainty in the assessment of doses to these organs. For a randomly chosen patient we therefore delineated the CTV, bowel cavity and rectum on all 30 CBCTs. The structures were copied to the planCT, where a fractional DVH was assessed for each of the structures. The DVH for each fraction were summed for each volume bin of 1 cm³ and the resulting DVH was compared to the corresponding DVH obtained from the planCT anatomy. An absolute volume scale was used since the CBCTs did not include the whole organ. The DVH sum was cropped at 60 cm³ because of the limited field-of-view of the CBCTs. The sum of the fractional DVHs gives an estimate of the upper limit of the delivered DVH. The volume of normal tissue irradiated was calculated for each fraction for the delivered ART course, as well as if the non-ART strategy had been applied.

Results

All patients completed the treatment without any breaks. The three treatment plans in the libraries were selected almost equally often, with a mean of 10, 9 and 11 for the small, the medium and the large size treatment plan, respectively (Figure 1). No clear pattern or time trend in the plan selections was observed. The time from CBCT acquisition to treatment was delivered was around eight minutes. All patients had systematic and daily feedback on which of the treatment plans that had been selected. None of the weekly CBCTs acquired for the first 10 patients indicated intra-fractional changes that would jeopardise target coverage. In 2% of the plan selection fractions the bladder extended outside the selected PSV, but in none of those fractions the bladder extended outside the treated PTV.

Figure 1. Plan selections performed for each of the 20 patients and 30 fractions. The first five fractions with large size plan correspond to the first week of treatment. Patient no. 12 had a pre-treatment boost of 10 Gy and plan selection was performed from first fraction in the ART treatment.

The mean ratio of the PTVs of the library plan relative to the CTV was 2.5, 3.1 and 5.0 for the small, medium and large size PTV, respectively (Figure 2). The course-averaged PTV for the actually delivered ART plans were reduced by 30% [11–54%] compared to non-ART. The course-averaged PTV was linearly dependent on the CTV (Figure 3): the linear regression analysis showed that use of ART reduced the course-averaged PTV with 183 cm³ [143–223 cm³] (R² = 0.94). For the bowel, the use of ART reduced the V45Gy by 100 cm³ and the V10Gy by 180 cm³ as compared to non-ART for patients treated to the bladder only (Figure 4). For patients receiving RT to both the bladder and the pelvic lymph nodes, the bowel sparing by ART was 100 cm³ in the high dose region (50–60 Gy). ART also lead to a considerable dose reduction for the rectum. The rectal V30Gy was reduced by 10% points in patients treated to the bladder only. As for the bowel, reduction was only observed at V45Gy or higher in patients who received RT to the pelvic lymph nodes.

Figure 2. Box plot of volumes for CTV, PTV_small , PTV_medium , PTV_large and PTV_course. The boxes are 25–75 percentiles with median as line inside the box, whereas upper and lower limits are the 95 and 5 percentiles.

Figure 3. Linear fit of PTV_course as a function of CTV for ART (black line) and non-ART (grey line). Assuming a spherical bladder shape, it can be shown also theoretically that the PTV is roughly a linear function of the CTV when the margin is much smaller than the radius of the sphere.

Figure 4. Population-average DVHs for bowel cavity (upper part) and rectum (lower part) for patients treated with RT to the bladder only (left) as well as patients where both the bladder and the pelvic lymph nodes were treated (right). The dotted lines represent one standard deviation and all curves are based on planCT data.

A large variation was observed in the daily CBCT-based fractional DVHs for the rectum for the case investigated (treated to the bladder only), with V40 Gy ranging between 0 and 30 cm³ (Figure 5). The DVH sum based on the daily CBCTs was higher than the DVH based on the planCT at the high dose levels. The median volume of normal tissue irradiated for this particular patient was 297 cm³ compared to 556 cm³ if the patient had been treated with non-ART (a reduction of 47%).

Figure 5. Rectal DVHs for one patient treated to the bladder only, showing the individual DVHs for each of the 30 treatment fraction with the adaptive scheme using anatomy from CBCTs (scaled to the full treatment dose), together with the average/summed DVH over each volume bin (dashed line) as well as the DVH for ART calculated on the planCT (solid black line). Doses below 10 Gy are not shown.

Discussion

In this first report of our clinical phase II trial of daily plan selection ART for bladder cancer, we found a median reduction in the course-averaged PTV of 30 %, ranging from 11% to 54% for the individual patients. In absolute volume, the reduction was 183 cm³ across the population. The sparing resulted in a decrease in dose to the bowel cavity and the rectum. In the patients who received RT to the pelvic lymph nodes there was less sparing of bowel cavity, and the rectum was only spared for doses above 45 Gy.

Different principles have been proposed for ART planning in treatment for bladder cancer. Meijer et al. reported clinical results from a study with ART planning based on two pre-therapy CT scans, one with full and one with empty bladder [24]. In their study ART was guided by Lipiodol markers which were used for demarcation of the tumour border [24–27]. Our findings, i.e. a reduction of treated volume of 30% are comparable with the findings of Foroudi et al. [28]. In the present study, a median reduction in course-averaged PTV was 183 cm³. Similar sparing was found by Tuomikoski et al. [29] and McDonald et al. [30], whereas the reduction of normal tissue irradiation was only 35 cm³ in a study by Burridge et al. [22]. The study by Burridge used plans with varying superior margin from CTV to PTV in combination with three-dimensional (3D) conformal RT. Since the whole bladder is considered as target, no dose reduction to the bladder was expected by use of ART. In case of partial bladder irradiation ART may allow for dose reduction to the bladder wall outside the PTV [31].

The approaches used to create the library plans differ between the previous studies [24,28–30,32]. In the present study we used the contours from four CBCTs acquired during the first week. The main advantage of this technique is that the CBCTs contain information about the real anatomical variations of the bladder, bowel and rectum. However, the library plans does not take into account the patterns of variation after the first week. For some patients the plan selection volumes closely encompasses the bladder as seen on the CBCTs, but two patients included in the study would have benefited from adaptation of the library plans because of a systematic change in bladder shape during the treatment course. Furthermore, one patient had a large bladder volume on the planCT and considerably smaller bladder size on the subsequent CBCTs (Figure 3). We recently investigated a strategy for ART planning using deformation vector fields resulting from deformable image registrations (DIRs) between CBCTs to the planCT [33]. In that simulation study we used the CBCTs from the first week of treatment, but based on automatic plan generation, the principle may be used throughout the whole treatment course to compensate for systematic anatomical changes. Anyhow, the daily plan selection ART technique as in the present study is a natural first ART technique to implement, and seems to give promising results. However, the method cannot be considered fully mature and further studies are needed to conclude on which ART strategy is optimal in different settings.

The aim of the present phase II trial is to reduce the rate of acute GI morbidity. The sparing of the bowel as reported in this study is therefore of key importance, as this has recently been shown to correlate with the risk of GI morbidity [34]. The bowel is, however, a highly mobile structure [35–37] and estimates not relying on daily dose recalculation are therefore prone to uncertainties. Dose accumulation is the gold standard for assessment of the delivered dose volume in both conventional and adaptive radiotherapy [38]. In the pelvic region, there are several limitations of the current commercial algorithms and to our knowledge there is no unsupervised DIR algorithm that is able to accurately map the random changes of rectum and bladder filling observed on daily CBCTs. One of the general problems is the trade-off in the regularisation of having invertible deformation fields versus enough deformation to cope with the changes [39–41]. Online re-optimisation, as described by Yan and co-workers, might be the ultimate ART principle [42]. However, here inaccuracies in the algorithms also pose problems. In a recent retrospective study, we found that for bladder cancer online re-optimisation could reduce the treated volume with a mean of 59% compared to non-ART method, i.e. a doubling of the volume reduction compared to the present study [43]. Due to the uncertainties related to dose accumulation, the present report focused on the volume reduction to be able to predict the reduction of treated volume prior to treatment delivery. However, DVH summation was performed for one patient, with the dose for each fraction and volume bin being summed to give an estimate of the delivered dose-volume relation. This may be considered as an upper limit to the dose delivered to the organs at risk.

In the post-fraction delivery CBCTs acquired weekly in our study, no deterioration in target coverage was indicated due to intra-fractional motion. Nevertheless, all intra-fractional motion cannot be assessed using CBCTs acquired after treatment, therefore cine-magnetic resonance imaging (MRI) scans were acquired weekly for the first 10 patients in the trial. These data are currently being analysed, but a preliminary analysis showed no deterioration of target coverage. Different values for intra-fractional margins have been suggested, with isotropic margins ranging from 2 mm to 7 mm in the Burridge, the present and the Foroudi studies and anisotropic margins in the Tuomikoski study [22,28,29]. The magnitude of the intra-fractional margins depends on the time from CBCT acquisition to treatment delivery completed and it is therefore dependent on the treatment technique (i.e. 3D conformal, IMRT or VMAT).

A logical next step to exploit the potential normal tissue sparing effect by ART would be to escalate the RT dose to the bladder tumour [26]. The use of a Lipiodol-based image-guided boost to the tumour only during the first 5 fractions combined with subsequent plan selection ART to the whole bladder in the remainder of the fractions might be an attractive strategy, leading to improved tumour control without increase of morbidity. In this approach the CBCTs acquired during the Lipiodol boost irradiation would be used for planning of the library of plans for the ART, i.e. the 5 fractions using large population-based margins would not be needed. Furthermore, the expected decrease in RT-related morbidity opens up for more aggressive chemotherapy. Organ preservation therapies for bladder cancer combine trans-urethral resection with chemo-irradiation. Results from recent studies indicate that bladder conserving strategies are promising and might be an alternative to cystectomy in selected patients with bladder cancer [1,44]. A randomised controlled trial comparing radical cystectomy to a chemo-irradiation trial (including state of the art image-guided ART) is warranted, however it may not be easy to conduct [45].

In conclusion, this study has showed that online ART for bladder cancer using daily selection of treatment plans constructed from delineations of CTVs on the CBCTs from first treatment fractions is feasible and leads to a considerable reduction of the course-averaged PTV. Final results of the trial have to be awaited to conclude whether this reduction translates into a clinically relevant decrease in radiation-induced side effects.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.