Dosimetric comparison between dynamic wave arc and co-planar volumetric modulated radiotherapy for locally advanced pancreatic cancer

Background: Dose reduction to the duodenum is important to decrease gastrointestinal toxicities in patients with locally advanced pancreatic cancer (LAPC) treated with denitive chemoradiotherapy. We aimed to compare dynamic wave arc (DWA), a volumetric-modulated beam delivery technique with simultaneous gantry/ring rotations passing the waved trajectories, with coplanar VMAT (co-VMAT) with respect to dose distributions in LAPC. Methods: DWA and co-VMAT plans were created for 13 patients with LAPC in the pancreatic head or body. The prescribed dose was 45.6 or 48 Gy in 15 fractions. The dose volume indices (DVIs) for the gross tumor volume, planning target volume (PTV), stomach, duodenum, small bowel, large bowel, kidney, liver, and spinal cord were compared between the corresponding plans. The values of the gamma passing rate, monitor unit (MU), and beam-on time were also compared. Results: The volumes of the duodenum receiving 39, 42, and 45 Gy were signicantly reduced to 2.8, 0.8 and 0.15 cm 3 in DWA from 3.9, 1.6 and 0.34 cm 3 in co-VMAT, respectively. The mean dose of the liver and D 2cm3 of the planning volume for the spinal cord were signicantly increased to 6.9 and 31.7 Gy in DWA from 5.9 and 30.2 Gy in co-VMAT, respectively. Meanwhile, there was no signicant difference in the target volumes except for dose irradiated to 2% of PTV (110.4% in DWA vs. 109.6% in co-VMAT). There were also no signicant differences in the other DVIs. Further, the gamma passing rate was 96.5% for DWA and 96.7% for co-VMAT (p = 0.65). The MU was signicantly higher in DWA than in co-VMAT (620 vs. 589, p = 0.001), and there was a signicant increase in the beam-on time (104 sec vs. 89 sec, p = 0.04). DWA was superior to co-VMAT regarding dose distributions in the duodenum in LAPC, albeit with slight increasing doses to the liver and the spinal cord and increasing MU and the beam delivery time. Further evaluation is needed to know how the dose differences would affect the clinical outcomes in chemoradiotherapy for LAPC.

D 2cm3 of the planning volume for the spinal cord were signi cantly increased to 6.9 and 31.7 Gy in DWA from 5.9 and 30.2 Gy in co-VMAT, respectively. Meanwhile, there was no signi cant difference in the target volumes except for dose irradiated to 2% of PTV (110.4% in DWA vs. 109.6% in co-VMAT). There were also no signi cant differences in the other DVIs. Further, the gamma passing rate was 96.5% for DWA and 96.7% for co-VMAT (p = 0.65). The MU was signi cantly higher in DWA than in co-VMAT (620 vs. 589, p = 0.001), and there was a signi cant increase in the beam-on time (104 sec vs. 89 sec, p = 0.04).
Conclusion: DWA was superior to co-VMAT regarding dose distributions in the duodenum in LAPC, albeit with slight increasing doses to the liver and the spinal cord and increasing MU and the beam delivery time. Further evaluation is needed to know how the dose differences would affect the clinical outcomes in chemoradiotherapy for LAPC.

Background
Pancreatic cancer has poor prognosis and is one of the leading causes of cancer-related deaths worldwide [1]. Surgical resection is the standard treatment for early stage pancreatic cancer, but 30-35% of patients are in a locally advanced stage upon diagnosis [2].These patients with locally advanced pancreatic cancer (LAPC) are commonly treated with systemic chemotherapy, but the median overall survival is only 11.9-13.6 months [3].A previous study on LAPC patients treated with rst-line A novel dose delivery method formed by simultaneous dynamic motion of the treatment couch and the linac gantry, which is called Trajectory-based Volumetric Modulated Arc Therapy (TVMAT), has been validated for use in stereotactic radiosurgery for the brain [13]. TVMAT produced comparable plans to VMAT in terms of the OAR doses while achieving better dose conformity and better OAR sparing compared to 3D-CRT plans. Another approach employing dynamic table and collimator rotations developing non-coplanar delivery technique called dynamic trajectory radiotherapy (DTRT) has been evaluated in various tumor sites including the prostate, lung, esophagus and head and neck [14]. DTRT plans could potentially spare OARs while maintaining or even improving the target coverage.
The dynamic wave arc (DWA) is a continuous non-coplanar delivery technique [15]. It allows simultaneous rotation of both radiation head unit and the O-ring-shaped gantry, thus delivering sequential non-coplanar beams both safely and quickly without the need to rotate the couch. The clinical bene t of DWA was rst reported in 15 patients with varying treatment sites, namely, breast boost, prostate, lung stereotactic body radiation therapy, and bone metastases [16]. Since then, DWA has been added to noncoplanar VMAT techniques to achieve better exibility in dose shaping while keeping dosimetrically effective delivery. DWA has been reported to have dosimetric advantages over coplanar VMAT (co-VMAT) in tumors located in the midline of the body, such as skull base tumors and prostate cancers [17,18]. A study on the application of DWA in cases of LAPC using preclinical versions of the treatment planning system (TPS) concluded that it has comparable dosimetric distributions to co-VMAT [19]. However, there are no previous reports focused on the dosimetric advantage of DWA in pancreatic head and body cancer, which is also located in the midline of the body, using the clinically integrated version of RayStation TPS (RaySearch Laboratories, Stockholm, Sweden). Thus, this study aimed to compare DWA with co-VMAT with respect to dose distributions in locally advanced pancreatic head and body cancer, using a TPS.

Study design and patients
This planning study evaluated 13 patients with LAPC in the pancreatic head (n=10) or body (n=3) who were treated with IMRT in our institution. The clinical stages (Union for International Cancer Control 7 th edition) were T3N0M0 in 5 patients, T4N0M0 in 7 patients, and T4N1M0 in 1 patient. The eligible criteria to this study are as follows: 1) primary tumors are located in the head or body of the pancreas, located close to the duodenum, 2) Planning tumor volume (PTV) sizes are less than 410 cm 3 and 3) PTV is included in the eld size of the Vero4DRT system (maximum, 15 cm × 15 cm).
Simulation and delineation of the target organ and organs at risk Patients were immobilized with a vacuum pillow (BodyFIX, Medical Intelligence, Schwabmuchen, Germany) in a supine position with both arms raised above the head. Expiratory breath-hold treatmentplanning CT was performed via an intravenous contrast-enhancing agent(without oral contrast agents) using either a LightSpeed RT scanner (GE Medical Systems, Waukesha, WI, USA) with slice thickness of 2.5 mm or SOMATOM De nition AS (Siemens Medical Systems, Erlangen, Germany) with slice thickness of 2 mm.
The gross tumor volume (GTV) included the tumor and metastatic lymph nodes. The clinical target volume (CTV) was de ned as the GTV plus a 5-mm margin as well as the retropancreatic space and the para-aortic lymph node region between 10 mm superior to the celiac axis and 10 mm inferior to the superior mesenteric artery. The (PTV) was de ned as CTV plus a 5-mm margin. The target organs and OARs were contoured based on the end-expiration phase CT. OARs were de ned for the stomach, duodenum, small bowel, large bowel, kidneys, liver, and spinal cord. The planning OAR volumes (PRVs) were created with a 5-mm margin for the duodenum, small bowel, and spinal cord and a 10-mm margin for the stomach.

Treatment planning
The co-VMAT and DWA plans were created for the Vero4DRT system (Hitachi, Tokyo, Japan) using RayStation version 6.2. Dose distribution was calculated using collapsed-cone convolution algorithm with a dose grid of 2.5 mm (LightSpeed RT scanner) or 2.0 mm (SOMATOM De nition AS). The prescribed dose was 48 Gy in 15 fractions to 95% of PTV boost volume, de ned as PTV excluding PRVs, while keeping the 36 Gy isodose line covering 98% of PTV volume. DWA treatment plans were created via synchronous rotation of the O-ring and the gantry with optimal sequencing of dynamic multileaf collimators. A single non-coplanar trajectory called pancreas peaks up trajectory ( Fig. 1) was applied to the DWA plans. The trajectory was divided into 5 segments. The O-ring rotated from 20° to 0°, 0° to 25°, 25° to 335°, 335° to 10°, and 10° to 340° during the rst, second, third, fourth, and fth segments, respectively. Simultaneously, the gantry rotated clockwise from 182° to 234°, 234° to 314°, 314° to 42°, 42° to 100°, and 100° to 178° in the ve segments, respectively. The co-VMAT treatment plans consisted of one arc that rotated clockwise from 181˚ to 179˚.
We applied the same prescription policy and the same constraints to both the DWA and co-VMAT plans.
Optimization was performed until the following criteria were met: (1) the maximum dose to the PTV was set to be <56 Gy, and (2) the dose to the OARs and PRVs should meet the dose constraints (Table 1) while keeping the dose coverage to PTV as mentioned above. If the PRV dose constraints cannot be ful lled, a 5% reduction of the prescribed dose (45.6 Gy to 95% of PTVboost-34.2 Gy to 98% of PTV) was allowed. The dose constraints were based on the previous papers from our department [7,20,21].

Patient-speci c quality assurance
The patient-speci c quality assurance of co-VMAT and DWA was performed using the ArcCHECK (Sun Nuclear Corp., Melbourne, FL, USA). The measured dose distribution was compared with the calculated results of treatment plans. The gamma analysis with global 3%/3-mm gamma criteria was conducted to compare the calculated and delivered dose distributions [22]. The gamma passing rate for areas receiving isodoses above 10% was calculated using a global difference approach for the absolute dose. In addition, data on the MU and beam-on time for co-VMAT plans and DWA plans were collected.

Plan evaluation and statistical analysis
The dose volume indices (DVIs) for the following parameters were compared between co-VMAT plans and DWA plans: dose irradiated to 98%, 95%, 50%, and 2% of GTV and PTV (D 98% , D 95% , D 50% , and D 2% ); irradiated volumes of the stomach, duodenum, large bowel, and small bowel to 39, 42, 45, and 48 Gy (V 39Gy , V 42Gy , V 45Gy , and V 48Gy ); mean dose (D mean ) and V 20Gy of the kidney; D mean of the liver; maximum dose (D max ) of the spinal cord; and near-maximum dose (D 2cm3 ) of the PRV for the spinal cord. The DVIs, gamma passing rate, MU values, and beam-on time were compared using the paired t-test on two paired samples. All statistical analyses were performed using R (version 3.5.1). A p value of <0.05 was considered statistically signi cant.

Dose-volume indices
The PTV ranged from 120.3 cm 3 to 406.1 cm 3 . There was no signi cant difference in GTV and PTV doses except for PTV D 2% between DWA and co-VMAT. PTV D 2% was signi cantly higher in DWA plans than that in co-VMAT, but the difference was only 1% (111.5±2.6% vs. 110.5±2.4%, p<0.01). The details for the target volume coverage are shown in Table 2. Figure 2 shows the dose distribution in a representative case using co-VMAT and DWA. For the OARs, DWA yielded signi cantly lower irradiated volume of the duodenum (V 39Gy , V 42Gy , and V 45Gy ) than co-VMAT (2.  Table 3). There was no signi cant difference in DVIs of the stomach, small bowel, large bowel, or kidneys between the two plans.

Discussion
Concurrent chemoradiotherapy has been shown to increase resectability and decrease postoperative recurrence in LAPC cases [6]. Pancreatic cancer is known to have fairly low radiosensitivity [23], and thus higher doses are needed for better local control. However, this also leads to a higher frequency of GI toxicities. The present study showed the dosimetric bene ts of DWA over co-VMAT in decreasing the intermediate and high doses to the duodenum, while preserving the PTV coverage. However, a trade-off relationship in irradiated doses among the OARs surrounding the pancreas was observed. With DWA, noncoplanar trajectories shift irradiation away from the duodenum toward other OARs namely the liver, spinal cord and left kidney. Therefore, the irradiated doses to the duodenum could be decreased with a trade-off of a slight increase in the mean liver dose and the maximum dose of the spinal cord. The other trade-off is the increase of the MUs followed by a prolongation of the treatment time.
Clinical trials assessing the bene ts of escalated dose radiation with an IMRT technique in LAPC have shown signi cantly increased local control and OS [7,24]. Goto  Gy). However, although dose-escalating IMRT trials have shown promising results, GI toxicities remain a major concern. Further reduction of GI toxicities is needed to adequately satisfy the need for dose escalation to achieve better local control.
Previous reports have showed the association between DVIs and GI toxicities, but the prescribed doses and fractionations varied among these reports [25-28]( Table 4). The BED is commonly used to compare such various doses and fractionation schedules with an alpha/beta ratio of 10 Gy (BED10) for acute toxicities and 3 Gy (BED3) for late toxicities in the duodenum. Considering the DVIs converted to BED10 and BED3, the DVIs for the duodenal toxicities proposed by the reports are consitent with most of the DWA plans in the present study (Table 4). Our data indicate that the use of DWA may be bene cial to decrease both acute and late duodenal toxicities, which is crucial for dose escalation in LAPC. However, it remains unclear whether the dosimetric superiority directly reduces GI toxicities because the toxicities are caused by multiple treatment-related factors such as chemotherapy and low platelet count [29,30].
Several methods of respiratory motion management have been investigated for the delivery of radiotherapy to pancreatic cancer patients. Abdominal compression, which reduces target motion by applying pressure to the patient's abdomen limiting diaphragmatic motion, is thought to cause pain and discomfort to the patient and alter the reproducibility of the OARs which may impact treatment delivery accuracy [31]. Breath-hold techniques can be used to reduce tumor motion during radiotherapy. The motion magnitude of the pancreatic head was found to be larger during inhalation breath-holds than during exhalation breath-holds [32]. Nakamura et al. have concluded that exhalation breath-hold is tolerable [33]. In their study, all the 10 tested pancreatic cancer patients succeeded in maintaining exhalation breath-holds for 20 seconds. In the present study, the breath-hold at the end of exhalation was assumed for both co-VMAT and DWA plans. DWA needed more MUs than co-VMAT (620 vs. 589), as well as prolongation of treatment time (104.1 sec vs. 88.8 sec). The 15-sec increase of the beam delivery time would lead to an increase in the number of the breath-holding sessions by one. The increase in MUs is presumably due to the increased beam path length because DWA irradiates a tumor from non-coplanar angles.
There are some limitations in the present study. First, the sample size is small, but it is still larger than that in a previous report on DWA that used a preclinical version of TPS [19]. Second, we focused on centrally located LAPC, and thus the bene t of DWA in tail pancreatic cancer or LAPC with larger PTV remains unknown. Given that this is a planning study, further investigation is warranted to explore the impact of increased liver and spinal cord doses on clinical outcomes, and for which cancer types and for which DWA trajectories can be optimal. Third, the present study evaluated only single-arc or singletrajectory plans for both DWA and co-VMAT. One reason for that is the shorter time for beam delivery preferable for breath-hold techniques. The other reason is that multiple coplanar arcs increase the tongueand-groove effect because the Vero4DRT system cannot rotate its collimator.

Conclusions
DWA was superior to co-VMAT regarding dose distributions in the duodenum in LAPC, albeit with slight increasing doses to the liver and the spinal cord and increasing MU and beam delivery time. Further evaluation is needed to know how the dose differences would affect clinical outcomes in chemoradiotherapy for LAPC. These ndings would help further the clinical studies of IMRT for LAPC, with escalating doses to the tumor while keeping less duodenal toxicities.

Declarations
Ethics approval and consent to participate This study was approved by the Institutional Review Board of Kyoto University Hospital (R1446) and was performed in accordance to the Declaration of Helsinki (1975, as revised in 2013). The need for written informed consent was waived.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
Takashi Mizowaki has an advisory contract with Hitachi, Ltd. Other authors declare that they have no con ict of interest.  Abbreviations: DWA = dynamic wave arc, co-VMAT = coplanar volumetric arc therapy, SD = standard deviation, GTV = gross tumor volume, PTV = planning target volume, and D x% = dose covered by the x%-isodose line Abbreviations are the same as in Tables 1 and 2.