Intra‐fractional patient motion when using the Qfix Encompass immobilization system during HyperArc treatment of patients with brain metastases

Abstract Purpose This study investigated the intra‐fractional motion (IM) of patients immobilized using the QFix Encompass Immobilization System during HyperArc (HA) treatment. Method HA treatment was performed on 89 patients immobilized using the Encompass. The IM during treatment (including megavoltage (MV) registration) was analyzed for six degrees of freedom including three axes of translation (anterior‐posterior, superior‐inferior (SI) and left‐right (LR)) and three axes of rotation (pitch, roll, and yaw). Then, the no corrected IM (IMNC) was retrospectively simulated (excluding MV registration) in three directions (SI, LR, and yaw). Finally, the correlation between the treatment time and the IM of the 3D vector was assessed. Results The average IM in terms of the absolute displacement were 0.3 mm (SI), 0.3 mm (LR) and 0.2° (yaw) for Stereotactic radiosurgery (SRS), and 0.3 mm (SI), 0.2 mm (LR), and 0.2° (yaw) for stereotactic radiotherapy (SRT). The absolute maximum values of IM were <1 mm along the SI and LR axes and <1° along the yaw axis. The absolute maximum displacements for IMNC were >1 mm along the SI and LR axes and >1° along the yaw axis. In the correlation between the treatment time and the IM, the r‐values were −0.025 and 0.027 for SRS and SRT respectively, along the axes of translation. For the axes of rotation, the r‐values were 0.012 and 0.206 for SRS and SRT, respectively. Conclusion Encompass provided patient immobilization with adequate accuracy during HA treatment. The absolute maximum displacement IM was less than IMNC along the translational/rotational axes, and no statistically significant relationship between the treatment time and the IM was observed.


| INTRODUCTION
Brain metastases are a common cause of morbidity and mortality in patients suffering from a variety of solid tumors and they affect 20-40% of cancer patients. As primary cancer management has improved, survival times have increased but so has the incidence of patients developing brain metastases. 1,2 Stereotactic radiosurgery (SRS) and stereotactic radiotherapy (SRT) deliver high doses of radiation to tumors in cranial lesions as a single fraction SRS or multiple fractions SRT and are gaining popularity for the treatment of brain metastases. SRS is now widely used for the treatment of patients with four or less brain metastases and a life expectancy >3-6 months. 3,4 However, Hughes et al. showed that the overall survival rate for patients with 5-15 brain metastases was similar to that of patients with 2-4 brain metastases when they received SRS. 5 Multi-isocenter irradiation techniques for multiple targets, such as Gamma Knife which is performed using a rigid and invasive fixed head ring, 6  provides single isocentric irradiation using a non-coplanar volumetric modulated arc therapy (VMAT) technique. This is combined with a simple treatment planning procedure that includes automated settings for the collimator angles, non-coplanar beam arrangement, and isocenter location. HA plans improve tumor conformity and reduce the dose of radiation applied to surrounding tissue. 8 Moreover HA requires a shorter treatment time than the conventional VMAT technique because it incorporates automated delivery (e.g., couch automation). Furthermore, for the treatment to be automated, the QFix Encompass TM immobilization system must be included in the HA treatment plan. The Encompass is a frameless-mask based system with a clam-shell style mask that was created by QFix (Avondale, PA, USA). 9 It allows the patient to be placed in an optimal position that ensures machine clearance during automated delivery.
It is necessary to set a small tumor margin for SRS and SRT treatment planning because the risk of radionecrosis increases with the gross tumor volume and despite local controls with a large margin there is no significant difference compared to a small margin. 10 A patient's position is decided in six axes using corn-beam computed tomography (CBCT) for image guided radiation therapy. However, SRS and SRT requires a relatively long treatment time (approximately 20 min) and intra-fractional patient motion (IM) may occur during dose delivery. IM may result in underdosing of the target or overdosing of the surrounding normal tissue. 11 Minniti et al. showed that IMs of up to 3 mm occurred when using frameless stereotactic systems. 6 However, the IMs when the Encompass is used during HA dose delivery have not been reported.
Therefore, this study aims to investigate IM during HA treatment when the patients are immobilized using the Encompass. Furthermore, images before, during, and after treatment will be analyzed to assess the necessity of monitoring patient motion.

2.B | Treatment planning
The CT images were loaded into a treatment planning system adding an additional 1 mm margin. The prescription dose was 20-24 Gy in a single fraction and 7-10 Gy in 3-5 fractions for 95% of the planning target volume for SRS and SRT, respectively. Each plan was designed for a 6-MV photon beam or flattening filter free beams with a 6-MV photon beam energy at maximum dose rates of 600 and 1400 monitor units per minute (MU/min). Overall, 3-4 arc fields were arranged (one coplanar arc with a 0°couch and noncoplanar arc fields at 315°, 45°, and 90°or 270°couch).

2.C | Treatment
Patients were immobilized using the Encompass as they were in the CT simulation. They were aligned to the isocenter location deter-

2.D | Data analysis
The IM during the HA treatment was defined as the difference between the patient's position before and after CBCT (including MV registration) in six dimensions (AP, SI, LR, pitch, roll and yaw).
Subsequently, the no corrected IM (IM NC ) was retrospectively simulated (excluding MV registration) in three dimensions (SI, LR, and yaw) The 3D IM was calculated as the square-root of the sum of squares of three translational/rotational IMs. The treatment time was defined as the time between the pre-and post-CBCT image acquisitions. The correlation between the treatment time and the 3D IM was assessed.
We used SPSS (version 24; IBM, USA) for the statistical analyses in this study. Initially, the Shapiro-Wilk test was performed to measure the normality of the distribution across all the IM and IM NC axes. In cases where the P < 0.05, measurement were non-normally distributed. In statistical comparisons between IM and IM NC , paired Wilcoxon signed-rank (non-normal distribution) and Welch (normal distribution) tests were used for the SI, LR, and yaw axes. In cases where the P < 0.05, we rejected the null hypothesis that there was no difference between IM and IM NC . In addition, Levene's test was performed to statistically assess the equality of variance between IM and IM NC along the SI, LR, and yaw axes. In cases where the P < 0.05, the variance was regarded as unequal. Furthermore, the   Table 1. The SD of IM NC F I G . 2. Flowchart of the HyperArc treatment process in this study. corn-beam computed tomography (CBCT) images were acquired before (pre-CBCT) and after (post-CBCT) dose delivery. A megavoltage (MV) image in the anterior-posterior or posterior-anterior direction was obtained during treatment and registered with the corresponding DRR image that was generated from the planning CT.

| RESULTS
was larger than that of IM for both SRS and SRT along the SI, LR, and yaw axes. There was a significant difference between IM and IM NC in SRS (P < 0.01) and SRT (P < 0.01) along the SI axes. There was no significant difference in the average IM along the LR and

| DISCUSSION
Historically, various types of masks have been used to effectively immobilize patients during SRS and SRT. Frame-mask based systems include rigid and invasive stereotactic head-rings and they are mainly limited to use with single-fraction treatments due to their invasive nature. 12 Consequently, frameless-mask based systems have grown in popularity since they are noninvasive, provide greater comfort for patients, and allow treatments to be fractionated while maintaining a high standard of immobilization. The Encompass is frameless-mask based system and an integral part of the HA high-definition radiotherapy automated SRS delivery workflow. 9 In this study, for absolute displacements using the Encompass during SRS, the average ± SD of IM for the translation axes were 0.3 ± 0.2 mm (AP), 0.3 ± 0.2 mm (SI) and 0.3 ± 0.2 mm (LR). Giuseppe et al. showed that, for absolute displacements using non-invasive relocatable frameless-mask based systems in SRS, the average ± SD of IM were approximately 0.1 ± 0.2 mm (AP), 0.1 ± 0.2 mm (SI) and 0 ± 0.1 mm (LR). 6 The IM were small in both their study and ours. Furthermore, Ramakrishna et al. demonstrated that a frameless-mask based system may provide better immobilization than invasive frame based mask systems when an orthogonal x-ray image-guidance system is used to correct for IM during treatment. 13 In our study, the MV image acquisition significantly reduces setup errors because the standard deviation of IM NC is larger than that of IM. Moreover the absolute maximum shifts of IM along the SI and LR axes were <1 mm in both SRS and SRT, but the absolute maximum displacement of IM NC along the SI and LR axes were 1.5 and 1.7 mm in SRS, and 1.6 and 1.2 mm in SRT. Therefore, it is necessary to correct for IM during HA treatment to compensate for doses with a small margin.
In this study, the average IM in terms of the absolute displace-

| CONCLUSION
This study demonstrated that Encompass provided patient immobilization with adequate accuracy during HA treatment with the absolute maximum displacements for IM <1 mm along the translational axes and <0.5°along the rotational axes. In addition, the average IM was less than IM NC along the translational/rotational axes, and no statistically significant relationship between the treatment time and the IM was observed. Consequently, we believe that MV image acquisition during treatment is useful for HA treatment.

ACKNOWLEDG MENTS
S.O. were involved in study design and data interpretation. All authors critically revised the report, commented on drafts of the manuscript, and approved the final report.

CONFLI CT OF INTEREST
No conflicts of interest.