Introduction
Set-up accuracy is important for patients receiving radiotherapy. Current techniques of image-guided radiotherapy such as radiotherapy using cone-beam computed tomography (CT) provide accurate set-up and high repeatability [1,2]. If there are remaining issues, the relative positions of organs and tumors may not be consistent at each time. Kidney cancer is usually located near the stomach, duodenum, small intestine, and colon. The relative positions of these radiosensitive organs and kidney cancer can be changed by various factors including peristalsis, accumulation of food or feces residue, intestinal gases, and level of muscular tension at the time of treatment. One solution for these issues is the use of magnetic resonance (MR)–guided online adaptive radiotherapy (MRgoART) with Unity MR-Linac (Elekta Unity, Elekta, Stockholm), which is equipped with a linear accelerator and a 1.5 Tesla magnetic resonance imaging (MRI) system [3]. In Unity MR-Linac, MR images can be used for radiotherapy image guidance: patients’ set-up and target shift (adapt-to-position method). The adapt-to-shape (ATS) method, which provides modification of tumor and organ delineation, can also be used with Unity MR-Linac. Optimization recalculation based on the modification is performed immediately before each fraction, and doses based on the recalculated plan are delivered [4]. In this way, inter-fractional relative position changes between kidney cancer and surrounding organs are modified, and more precise radiotherapy can be delivered than adapt-to-position method. Here, a new question arises regarding MRgoART: when kidney cancer is located adjacent to the small bowel, leading to a disturbance of sufficient radiotherapy dose delivery, could this situation be changed on another day? The timing of irradiation might be more important in hypofractionated radiotherapy especially in single-fraction radiotherapy (stereotactic radiosurgery [SRS]) which has been reported to provide better local control than multifraction radiotherapy [5]. In this planning study, a kidney lesion located near the gastrointestinal tract was simulated for SRS using the ATS method. The purpose of this study was to estimate the delivery dose to targets at each MRgoART timing and to investigate the values of waiting for good timing to perform SRS.
Materials and Methods
1. Patient and tumor selection
Patients who received stereotactic radiotherapy (SRT) for primary kidney cancer using the 1.5 Tesla Unity MR-Linac system between February 2022 and February 2024 at Tohoku University Hospital were identified from our database. Targeted tumors that were located within 1 cm from the organs at risk (OARs) were selected. In this study, the OARs included the stomach, duodenum, small bowel, and colon (gastrointestinal tract). Other non-irradiated tumors such as cystic lesions located within 1 cm from the OARs were identified with a limitation of one tumor per kidney. A flowchart of the selection for this study is shown in Supplementary Fig. S1. All eligible tumors were replanned for SRS and analyzed.
2. Set-up and SRT procedure in actual treatment
Patients were placed in the supine position with or without a waist belt for abdominal compression (Beruetto, Taketora, Kanagawa, Japan). At the time of pretreatment simulation, both MRI and CT scans were performed for radiotherapy planning. At each treatment, patients were immobilized at the same position as that for pretreatment planning, and then a T2-weighted MRI for MRgoART was obtained. The pretreatment contours were adapted by deformable image registration onto the online planning MRI and the contours were manually modified. Then, optimization recalculation was performed, checked, verified, and finally the beam was delivered. Our treatment workflow of Unity MR-Linac was described elsewhere [6]. The step-and-shoot method of intensity-modulated radiotherapy with 7 MV flattening-filter-free photons was used for the radiotherapy treatment plan. The interval between each fraction was at least 40 hours, and the overall treatment time was within 14 days.
3. Contouring and planning simulation in this study
Contouring and planning were performed in Monaco v5.51.11 (Elekta, Stockholm). The eligible tumors, including both the irradiated tumors and non-irradiated tumors, were contoured as exhalation phase gross tumor volume (GTV) on a T2-weighted planning MR image of the exhalation phase using a breathing navigator [7]. The inhalation phase GTV was also contoured from the inhalation phase of T2-weighted planning MR image. Then the internal target volume (ITV) was created from the exhalation phase GTV by adding minimum asymmetric margins to fully cover the inhalation phase GTV. The planning target volume (PTV) was created from the ITV plus 0.5 cm for all directions to account for inter-fractional tumor motion variation and intra-fractional respiratory and positioning uncertainties. The PTV was not trimmed in this study to facilitate easier comparison of the dose coverage change of the PTV although some institutions report trimming the PTV to avoid overlap with the OAR [8].
When SRT with the ATS method was used, only T2-weighted MR image of the exhalation phase was scanned. The exhalation phase GTV was recontoured on this MRI, and then the ITV was created from the exhalation phase GTV by adding the same margin as that in the pretreatment planning MRI. The ITV plus 0.5 cm margin was used to define the PTV. The PTV was used for motion monitoring, and the GTV was assessed to ensure it remained within the PTV during respiratory motion on the treatment day using cine MRI [6]. If the GTV moved beyond the PTV, the abdominal belt was retightened. In cases where the GTV still exceeded the PTV despite retightening the belt, the PTV margin was expanded beyond 0.5 cm. Following this, the OAR was recontoured, and inter-fractional changes and movements of the OAR were modified. However, additional margins for intra-fractional changes were not applied to the OARs, based on a prospective study in which planning organ-at-risk volume margins were not mandatory, even though the ATS method was not used [9].
In this study, SRT was planned as single-fraction radiotherapy, i.e., SRS. Twenty-six Gy was delivered to cover the highest percentage, up to 99%, of the PTV, and 32.5 Gy was delivered to 0.03 cm3 of the PTV (80% isodose), adhering strictly to the upper limit of the OAR dose constraints (Table 1). The highest percentage of PTV coverage by the prescribed dose of each plan was investigated. The prescribed dose and dose constraints were based on the protocol of clinical trial (TROG 15.03) and other well-known constraints [9,10]. We created 18 planning series: 18 pretreatment simulation plans and 54 MRgoART plans (3 MRgoART plans per 1 tumor). The MRgoART plan was created as three separate SRS plans obtained from the 3-fraction MRgoART for kidney cancer. The ATS method was used in all MRgoART plans. In all pretreatment simulation plans and MRgoART plans, dose-volume data were evaluated using the following parameters: the percentage of the PTV covered by prescribed dose (26 Gy), the minimum dose covering at least 99% of the PTV, ITV, and GTV, mean doses of the PTV, ITV and GTV, the minimum dose covering at least 0.03 cm3 of the PTV, ITV and GTV, prescription isodose volume (PIV) which is the volume covered by the 26 Gy isodose line, prescription isodose target volume (PITV) which is the target volume covered by the 26 Gy isodose line, conformity index (CI) which is calculated as PITV/PIV, Paddick conformity index (PCI) which is calculated as (PITV)2/(PTV×PIV), homogeneity index (HI) which is calculated as (D2%−D98%)/D50% where Dx% is the minimum dose covering at least x% of the PTV, and OAR doses [11-13]. The best and worst MRgoART plan were defined as the plan that achieved the highest and the lowest percentage of the PTV coverage of the prescribed dose. We also evaluated the magnitude of overlap between the PTV and gastrointestinal tract, as well as between the PTV plus 0.5 cm and the gastrointestinal tract, in both pretreatment and MRgoART plans.
4. Statistical and simulated clinical treatment analysis
A two-sided Wilcoxon signed-rank sum test was used to compare pretreatment simulation plans and MRgoART plans. The Pearson’s correlation coefficients were calculated to measure the strength and direction of the relationship between two factors. A p-value less than 0.05 was defined as significant. Statistical analyses were performed using EZR version 1.54 (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a modified version of R commander (R Foundation for Statistical Computing, Vienna, Austria) [14].
To reveal the effectiveness of waiting for good timing to perform SRS in a clinical setting, clinical scenario simulation which was the evaluation of MRgoART plan in the order of actual treatment was performed. In the clinical MRgoART simulation, each plan selected the maximum coverage (up to 99%) of the PTV within the dose constraints of the OARs and a change in the PTV coverage by 5% was used as a cut-off. This is because although the TROG 15.03 trial required 99% prescribed dose coverage of the PTV, a PTV coverage of 95% (within a 4% change) was acceptable to fulfill the dose constraints of the OAR [9]. If the PTV coverage decreased from 99% to 94% (5% decrease), the plan would be considered the protocol violation. Therefore, this simulation used the 5% difference as the cut-off. If the prescribed dose (26 Gy) covered 95% or more of the PTV within the dose constraints, the plan was acceptable and SRS would be performed [9]. If the prescribed dose coverage of the PTV improved by 5% or more from pretreatment plans within the dose constraints, the plan was also regarded as acceptable and SRS would be performed. MRgoART timing of these two plans was regarded as good timing. On the other hand, the timings of other MRgoART plans in which the change of PTV coverage of the prescribed dose was within 5% (plan with almost no difference) or decreased by 5% or more to fulfill the dose constraints were regarded as fair timing and bad timing, respectively. MRgoART plans of these two timings were unacceptable, and they proceeded to the next round of MRgoART. Similarly, the MRgoART plans for the remaining tumors were estimated at the second and third MRgoART timings. Finally, all remaining patients would be treated at the third adaptive radiotherapy (ART).
Results
Eighteen tumors met the inclusion criteria: five tumors were actually irradiated and 13 tumors were not irradiated (Supplementary Fig. S1). The characteristics of 14 patients and 18 tumors used for analysis in this study are summarized in Table 2. The median overall treatment time was 8 days (range, 7 to 8) and all intervals between each fraction were 40 hours or more. There was no patient who needed the PTV margin beyond 0.5 cm on the treatment day. The results of pretreatment simulation planning and MRgoART planning using ATS are summarized in Table 3. Among these factors, Spearman correlation coefficients between the percentage of PTV covered by the prescribed dose and PIV, PITV, CI, PCI, HI, the overlap between the PTV and gastrointestinal tract, the overlap between the PTV plus 0.5 cm and gastrointestinal tract, and the minimum dose covering at least 1.5 cm3 of the ipsilateral kidney were 0.48 (p < 0.001), 0.44 (p < 0.001), 0.10 (p = 0.378), 0.88 (p < 0.001), −0.71 (p < 0.001), −0.27 (p = 0.020), −0.35 (p = 0.002), 0.77 (p < 0.001), respectively. Among the 18 planning series, 4 planning series achieved 100% dose coverage of 99% of the PTV through pretreatment planning and MRgoART. Other series included at least one plan that required a reduction of the prescribed dose coverage of the PTV to fulfill the dose constraints of the OAR. Among these plans that did not achieve 100% dose coverage of 99% of the PTV, the dose coverage limiting OARs were the stomach in one plan, the duodenum in 14 plans, the small bowel in 29 plans, and the colon in four plans. The prescribed dose moderately or greatly differed at the MRgoART timing, as shown in Table 3. Some examples are shown in Supplementary Figs. S2 and S3. In Supplementary Fig. S2, no overlap between PTV and OARs was observed on the day (A and B), however, some overlap between the PTV and small bowel emerged while the duodenum moved away from PTV on another day (C and D). Supplementary Fig. S3 is an example of almost the same axial slice on different days (A, B, C, and D). The magnitude of overlap between PTV and small bowel and overlap between PTV and colon were different on each day, therefore, dose-limiting OARs changed. In addition to organ movement, gastrointestinal delineation was deformed by accumulation of food or feces residue.
The individual data for the percentage of covering the PTV by 100% of the prescribed dose (26 Gy) at pretreatment and three times of MRgoART are shown in Fig. 1. The difference between pretreatment simulation plans and MRgoART plans ranged from a 78% increase to a 92% decrease, with a mean ± standard deviation of −2.7% ± 26.5%. Among the MRgoART SRS plans, the best and worst MRgoART plans are summarized in Table 4. A comparison of pretreatment simulation plans and the best MRgoART plans showed significant changes in all of the parameters of PTV, ITV, and GTV. On the other hand, a comparison of pretreatment simulation plans and the worst ART plans showed no significant change in any of the parameters. There were only six of the 18 planning series (33%) that consistently showed change in PTV coverage of the prescribed dose within 5% at all timings.
To estimate the optimal timing to perform SRS, the MRgoART plans were investigated in the order of actual treatment (Fig. 2). Seven of the pretreatment simulation plans showed 95% or more of the PTV covered by the prescribed dose, and these plans would fall into only two categories at MRgoART: achieving 95%–99% PTV coverage by the prescribed dose or experiencing a decline of 5% or more in PTV coverage by the prescribed dose. On the other hand, one plan showed 5% or less PTV coverage by the prescribed dose and this plan would never show a decline of 5% or more in PTV coverage by the prescribed dose. At the first MRgoART, six of the 18 plans achieved 95% or more PTV coverage by the prescribed dose. These six plans consisted of four plans with 99% PTV coverage at both the pretreatment timing and the first MRgoART timing, and the other two plans showed an increase from less than 95% PTV coverage at pretreatment simulation. Additionally, two plans and one plan showed 95% or more PTV coverage by the prescribed dose at the second MRgoART timing and third MRgoART timing, respectively. Two plans achieved more than 5% increase from the pretreatment simulation plan at the first MRgoART timing, and two plans and three plans achieved more than 5% increase at the second MRgoART timing and MRgoART timing, respectively. At the last chance of MRgoART, there was one tumor that showed almost no difference in the plan from the pretreatment simulation plan and there was only one tumor that showed a more than 5% decrease in the prescribed dose coverage of PTV from the pretreatment simulation plan. In summary, 16 of the 18 planning series would be irradiated at a good timing within the three-time chance of MRgoART, whereas nine of the 18 planning series would be irradiated at a good timing at first MRgoART.
Discussion and Conclusion
Although this study was a planning study, to the best of our knowledge, this study is the first study in which dose coverage differences at MRgoART timing was investigated. MRgoART has been used for many sites and many kinds of cancers including the location that was close to OAR [15-20]. In this study, the best benefit-risk balance of kidney tumors located close to the OAR differed depending on the MRgoART timing (Fig. 1). If the benefit-risk balance leans towards risks, radiation oncologists might postpone performing radiotherapy, especially SRS, with the expectation that the next chance of MRgoART will show a better balance. This expectation and waiting for next timing is possibly one of the options considering the result of Fig. 1. Because there was no guidance to irradiate or to wait, the clinical treatment scenario simulation was investigated using a 5% difference as the cut-off (Fig. 2). Although we investigated only three timings for SRS, the result of the clinical scenario simulation would be promising. Among the 18 tumors, 16 tumors would be irradiated at good timing, one tumor would be irradiated at fair timing and one tumor would be irradiated at bad timing using the three timings, in comparison to five tumors irradiated at fair timing and five tumors irradiated at bad timing at the first ART. These results encourage the expectations of radiation oncologists that the next timing of MRgoART will show a better benefit-risk balance, with the caution that each timing involved non-consecutive days and that does not guarantee the future outcome. It was also noted that the good timing was not always a good candidate for SRS. In Fig. 1, patients number 6 and 9 had consistently low percentages of PTV coverage by prescribed dose and were not considered good candidates for SRS. Another method which improve the benefit-risk balance is to reduce the ITV and PTV margins. The future respiratory gating system (known as Comprehensive Motion Management in Elekta Unity, Elekta, Stockholm) will provide a better benefit-risk balance.
SRS for kidney cancer has been reported to have a better local control rate than multifraction radiotherapy and has shown good local control in prospective studies [5,21]. Multifraction radiotherapy, such as a 10-fraction schedule, is an option to deliver a higher total dose to the target when kidney cancer close to OAR [22]. However, there are some difficulties, especially in terms of machine time of the Unity MR-Linac. MRgoART with the ATS method takes a long treatment time, with a median time of 42 minutes and a wide range from 30 to 65 minutes [23]. Additionally, MRI for pretreatment radiotherapy planning was often performed using the Unity MR-Linac system and it took further machine time. The disadvantage can affect the choice of multifraction radiotherapy. In our clinical scenario simulation, if SRS was selected, nine of the 18 tumors and six of the 18 tumors required the second MRgoART timing and third MRgoART timing, respectively. Although there is no data showing that SRS for kidney cancer with timing strategy is equivalent to multifraction radiotherapy, it can help to reduce machine time on the Unity MR-Linac.
MRgoART of Unity MR-Linac also has an adapt-to-position method, which provides only a shift of the plan to the patient. The adapt-to-position method is relatively simple and therefore it takes a shorter time than that for the ATS method [23]. However, the ATS method has the advantage of modifications of the relative positions of organs and tumors that cannot be made by using the adapt-to-position method of MR-Linac or by using the method for shift and rotation of patients in the most modern Linac [4]. The ATS method has been shown to be more effective than the adapt-to-position method for prostate cancer and head-and-neck cancers [24,25]. When kidney tumors are adjacent to the OARs, the effectiveness of the ATS method will be enhanced, and the ATS method will offer the best benefit-risk balance immediately before irradiation. It can be deduced from the results of this study that only six of the 18 plan series consistently showed that the change in PTV coverage of the prescribed dose was within 5% at all timings, and the other 12 plan series potentially included adaptive plans of underdose PTV or overdose OAR by the adapt-to-position method. Therefore, the ATS method is desirable for kidney tumors when they are adjacent to the OAR.
A comparison of the pretreatment simulation plans with the best MRgoART plans or the worst MRgoART plans showed a significant difference only between the pretreatment simulation plans and the best plans. This was unexpected to some extent because seven of the 18 plans showed 95% or more PTV coverage by the prescribed dose at pretreatment simulation, which were difficult to improve the coverage in the MRgoART plan. On the other hand, only one plan showed 5% or less PTV covered by the prescribed dose at pretreatment simulation, which hardly worsened the coverage in the MRgoART plan. Although there was such an unexpected result, the result would encourage expecting to obtain the best MRgoART plan at another timing.
There are some limitations in this study. Firstly, this was a planning simulation study using retrospective data, and some of the eligible tumors in this study were not actually irradiated. In ATS, only T2-weighted MR images of the exhalation phase were scanned, and therefore, inter-fractional changes in the ITV were not considered. In the clinical scenario simulation, optimal timing was assessed using retrospective data without consideration of future MRI data, which compromises the validity of the clinical scenario. In addition, a change of 5% was used as a cut-off value; however, the 5% value was not based on clinical evidence and the optimal cut-off value was not investigated. Similarly, the three timing assessments were used for each patient in this study, but how long one should wait for the optimal timing was not investigated. Furthermore, the clinical effectiveness of a PTV coverage of less than 95% remains unclear. Therefore, the results of this study are not always applicable to clinical treatment, and further evidence is needed.
In conclusion, MRgoART revealed that target coverage by the prescribed dose and the dose received by the target were different at each MRgoART timing, and the difference between the pretreatment simulation and the best MRgoART timing was significant. In a clinical scenario simulation, waiting for a good irradiation timing might be one of the option in case of suboptimal timing.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
