Injection of hydrogel spacer increased intrafractional prostate motion during volumetric modulated arc therapy-stereotactic body radiation therapy for prostate cancer

The aim of this study was to clarify the association between intrafractional prostate shift and hydrogel spacer. In order to move the the injection proposed to the patients as an option in case of meeting the indication of use. We monitored intrafractional prostate motion by using a 4-dimensional (4D) transperineal ultrasound (US) device: the Clarity 4D ultrasound system (Elekta AB). The deviation of the prostate was monitored in each direction: superior-inferior (SI), left-right (LR), and anterior-posterior (AP). We also calculated the vector length (VL). The maximum intrafractional displacement (MID) per fraction for each direction was detected and mean of MIDs was calculated per patient. The MIDs in the non-spacer group and the spacer group were compared using the unpaired t-test. irradiation.


Background
External beam radiation therapy (EBRT) is recognized as one of the primary treatment options for patients with prostate cancer (PCa) (1,2). Intensity modulated radiation therapy (IMRT) with imageguided radiation therapy (IGRT) technique is currently the gold standard for EBRT. A low α/β ratio of PCa has encouraged hypofractionation and stereotactic body radiation therapy (SBRT), or extreme hypofractionation, which is currently considered as a promising option of EBRT (3,4). While higher quality of positioning is required in SBRT to optimize treatments, it is known that organs in pelvis including prostate are shifting under the in uence of rectal volume, bladder volume, and change of muscle tension among other things (5)(6)(7)(8)(9)(10)(11)(12). These intrafractional uncertainties possibly affect the dose distributions (13,14).
Less is known about risk factors related to intrafractional prostate motion. While the variability in location of external skin markers relative to internal anatomy in obese patients caused a signi cant difference in terms of interfractional prostate shift (15), there was no relationship between body mass index (BMI) and intrafractional prostate motion (16). It was reported that the shorter the maximum rectal diameter is, the less the intrafractional prostate motion is (17).
The injection of hydrogel spacer between the prostate and the rectum has been introduced for RT for PCa to separate the prostate from the anterior wall of the rectum, which contributes to reducing the RT dose of the rectum. It was shown that the insertion of hydrogel spacer did not greatly limit interfractional and intrafractional prostate displacements (18)(19)(20). We hypothesized that the hydrogel spacer potentially implicates the prostate variability during RT. The aim of this study was to clarify the association between the intrafractional prostate shift and the hydrogel spacer.

Methods
A total of 38 patients with histologically con rmed prostate cancer who received de nitive volumetric modulated arc therapy (VMAT)-SBRT with prostate motion monitoring with or without androgen deprivation therapy (ADT) for PCa in our institution in 2018-2019 were retrospectively evaluated. The study was reviewed and approved by the institutional review board and ethics committee. Examination number was 3372.

Radiotherapy
A total dose of 36.25-40 Gy in ve fractions was prescribed to 95% of the planning target volume (PTV) every other weekday. All patients received CT scans which were reconstructed 1-mm-thick slices with a full bladder for treatment planning. A rectal enema was prescribed before simulation and before each treatment session to empty the contents of the rectum. The clinical target volume (CTV) consisted of prostate with or without seminal vesicles according to the risk classi cation of the NCCN guidelines version 1.2018. The CTV was extended by 5mm in every direction except posterior with 3mm extension to generate the PTV. We used Monaco (Elekta AB, Stockholm, Sweden) as the treatment planning system. In order to move the rectum away from the prostate, the hydrogel spacer (SpaceOAR system, Boston Scienti c, Marlborough, the United States) was transperineally injected into the recto-prostatic space of patients who hoped to receive it in case of meeting the indication of use. KV cone beam CT (CBCT) scans were acquired after the setup before each treatment session to reduce the interfraction error of patient positioning. motion monitoring Intrafraction motion of the prostate was monitored by a 4-dimensional (4D) transperineal ultrasound (US) device: the Clarity 4D ultrasound system (Elekta AB) with an autoscanning perineal US probe. We regarded the prostate position when CBCT scans nished as the baseline position. Monitoring time was de ned as the time from the end of CBCT to the end of radiation. The deviation of the prostate from the baseline position was monitored as a function of time along the three directions: superior-inferior (SI), left-right (LR), and anterior-posterior (AP). We also calculated the vector length (VL) at each point in time.
Large spike-like prostate displacement was seen during couch shift for patient position adjustment in some fractions (Fig. 1) though the prostate position we evaluated was relative to the couch position. We excluded these displacements from analyses because all patients immediately recovered from this error before the start of radiation. The maximum intrafractional displacement (MID) per fraction for each direction was detected and mean of MIDs was calculated per patient. The Stroom formula (21) (= 2.0 Σ + 0.7 σ) and the van Herk formula (22) (= 2.5 Σ + 0.7 σ) were used to calculate the CTV-PTV margin derived from intrafractional error, where Σ is the systematic error and σ is the random error. statistical analysis Differences in characteristic variables between patients without spacer and those with spacer were tested using the χ2-test for categorical variables and the unpaired t-test for continuous variables. In this study, the volume from the slice 1 cm above the highest part of the PTV to the slice 1 cm below the lowest part of the PTV in axial slices was considered as rectum volume. Bladder volume in this context included internal urine. The MIDs in the non-spacer group and the spacer group were compared using the unpaired t-test. Multiple regression analysis was used to detect risk factors related to intrafractional prostate motion. All statistical analyses were two-sided and performed using R, version 4.0.3. Results were considered statistically signi cant at P < 0.05.

patient characteristics
We reviewed 33 fractions in eight patients as the spacer group and 148 fractions in 30 patients as the non-spacer group. The baseline patient characteristics were presented in Table 1. There was no statistically signi cant difference between the two groups except for rectal volume. The median age was 73 (58-85) and 79 (66-84) years (P = 0.22) and the mean monitoring time was 275 ± 42 (mean ± SD) and 279 ± 33 seconds (P = 0.82) in the non-spacer group and the spacer group, respectively. Twenty-one patients (70.0%) in the non-spacer group and six patients (75.0%) in the spacer group received neoadjuvant ADT (P = 0.99), respectively. The mean BMI was 23.5 ± 2.7 and 23.6 ± 3.4 kg/m 2 in the nonspacer group and the spacer group, respectively (P = 0.93). Nine patients (30.0%) in the non-spacer and one patient (12.5%) in the spacer group underwent abdominal surgery before RT, respectively (P = 0.65). The mean volume of prostate (non-spacer vs. spacer group, 31.1 ± 12.4 cc vs. 25.0 ± 9.2 cc, P = 0.22) and bladder (277.1 ± 169.2 cc vs. 205.2 ± 77.9 cc, P = 0.26) measured on planning CT in each group was not signi cantly different. The rectum volume on planning CT in the non-spacer group was larger than the spacer group (55.4 ± 15.5 cc vs. 43.4 ± 9.0 cc, P = 0.047). The comparison of MID for each direction and the max of intrafractional VL between the two groups was presented in Table 2 and the boxplots of them were shown in Fig. 2 The superior and anterior MIDs were smaller in the non-spacer group, while the other MIDs and VL were not signi cantly different in the two groups. The results of multivariate analyses for MIDs for each direction were shown in Table 3. We included age, spacer injection, rectum volume, and the duration of monitoring as explanatory variables. Spacer injection was the independent risk factor of superior and anterior MIDs. There was no independent risk factor of inferior, left, right, and posterior MIDs and maximum VL. They reported that the displacements of prostate (mean ± SD) were 0.0 ± 3.4 mm, 0.0 ± 1.5 mm, 0.2 ± 2.9 mm in the SI, LR, and AP dimensions, respectively. Willoughby et al. (9) used the Calypso 4D localization system which is real-time tracking system with implanted electromagnetic transponders to track the intrafractional shift of prostate. They showed that the average (± SD) of the maximum differences in 11 cases were 3.61 ± 3.13 mm, 0.91 ± 0.35 mm, 3.92 ± 4.32 mm in the SI, LR, and AP directions, respectively.
Pinkawa et al. (11) demonstrated that the intrafractional displacements of prostate (mean ± SD) were 0.0 ± 2.0 mm, 0.2 ± 1.9 mm, 0.6 ± 2.2 mm in the SI, LR, and AP directions in 32 patients with prostate cancer by using transabdominal US tracking system. Comparable level of the intrafractional prostate motion with these studies was seen in our study. The average (± SE) of the maximum vector displacement was 2.24 ± 0.19 mm and 2.89 ± 0.62 mm in the non-spacer and the spacer group, respectively.
Shihono et al. (12) suggested the patient population-based margin according to the van Herk formula is as follows: 1.10 mm, 1.25 mm, and 1.33 mm in the SI, LR, and AP directions, respectively. They used the Clarity system just like our study. We demonstrated the larger margins calculated based on our population; 3.14 mm, 2.81 mm, 4.23mm in the non-spacer group and 4.27 mm, 3.37 mm, 5.00 mm in the spacer group. The difference is probably ascribed to the fact that Shihono et al. may have used mean intrafractional motion for margin calculation, whereas we used maximum intrafractional motion.
Knowledge about parameters related to intrafractional prostate motion is absolutely limited. Brown et al. (16) showed that there was no statistically signi cant relationship between intrafractional prostate motion and BMI by using linear regression analysis. Oates et al. (17) investigated a relationship between maximum rectal diameter (MRD) and intrafractional prostate motion. They showed with 90% con dence that for a MRD ≤ 3 cm, prostate displacement will be ≤ 5 mm and that for a MRD ≤ 3.5 cm, prostate displacement will be ≤ 5.5 mm. By prescribing a rectal enema and performing CBCT before each treatment session, the variety of MRD may have been minimized in our study. Rectum volume was smaller in the spacer group, which may be caused by the deformation of rectum by the pressure from the anterior direction by the injected hydrogel spacer (Fig. 3). However, rectum volume was not an independent risk factor for prostate displacement in the multivariate analysis. The displacement of prostate was shown to be smaller in step-and-shoot IMRT fractions than in VMAT fractions due to the shorter treatment time of VMAT by Ballhausen et al. (24). In the present study, we treated all patients with VMAT using attening lter free (FFF) beams and monitoring time from the end of CBCT to the end of radiation was about 4.5 minutes. According to our study, monitoring time did not signi cantly affect prostate shift.
Picardi et al. (19) showed that hydrogel spacer injection into the recto-prostatic space did not signi cantly in uence the interfraction prostate motion based on the analysis using implanted ducial markers and CBCT. It was reported that hydrogel spacer insertion signi cantly reduced the intrafraction rotational shift in the AP direction on cine-MRI by Cuccia et al. (20) and they concluded that hydrogel spacer contributed to limiting prostate intrafractional motion. On the other hand, Juneja et al. (18) showed that the average of the mean intrafractional vector displacement of prostate was signi cantly larger in patients with hydrogel spacer than those without spacer by analyzing the implanted electromagnetic markers position on kV uoroscopy. The difference between the two groups was 0.4 mm on their study. In our study, there was no signi cant difference in maximum VL, whereas superior and anterior MIDs were signi cantly larger in the spacer group in our study, and the difference between the two groups were 0.5 mm in the superior direction and 0.7 mm in the anterior direction.
The limitation of our study was the fact that the quality of our results depends on the accuracy of the

Consent for publication
Patients gave written consent for data collection and analysis.

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
The authors declare that they have no competing interests in association with this study.

Funding
Not applicable.

Authors' Contributions
SS collected and assembled the data, drafted the manuscript and critically revised the article for important intellectual content. HY supervised all of the above work. MO, YN, YW, and TI helped to interpret the collected data. OA helped to draft the manuscript. All authors read and approved the nal manuscript. An example of prostate motion in the superior-inferior direction during single fraction. Red rectangular area means the duration of couch shift. Large spike-like displacement was seen while the couch was shifting.