Evaluating the positional uncertainty of intrafraction, adjacent fields, and daily setup with the BrainLAB ExacTrac system in patients who are receiving craniospinal irradiation

Abstract Purpose To investigate the daily setup, interfraction motion, variability in the junction areas, and dosimetric effect in craniospinal irradiation (CSI) patients. Methods Fifteen CSI patients who had undergone split‐field IMRT were followed in the study. Previous, middle, and posttreatment, each target volume position was evaluated using the ExacTrac system. Interfraction and intrafraction motions, the margin of the junction in adjacent targets volumes, and the dosimetric effect of the longitudinal residual error were analyzed. Results The lowest attainment rate within the tolerance of the initial setup error was 66.79% in six directions. The values of the initial error were within 15 mm (SD 4.5 mm) in the translation direction and 5° (SD 1.3°) in the rotation direction after the transposition of the treatment isocenter. With the guidance of the ExacTrac system, the interfraction and intrafraction residual errors were almost within the tolerance after correction, the margin of CTV‐to‐PCTV was in the range of target expansion criteria. The residual longitudinal errors resulted in only slight changes in the mean doses of PGTV and PCTV, while the maximum dose of the spinal cord increased by 16.1%. The patients did not exhibit any side‐effects by the overall treatment during the follow‐up period. Conclusions Position correction is necessary after setup and the transposition of the treatment isocenter. Intra‐fraction motion in the lateral direction should be monitored throughout treatment. The position errors in junction areas are almost within the tolerance after correction. The patients did not exhibit any side‐effects by the overall treatment.


| INTRODUCTION
Radiotherapy (RT) or adjuvant RT following surgery is the current standard treatment for patients with an intracranial germinoma and medulloblastoma, which has dramatically increased the 5-year survival rate; Craniospinal irradiation therapy (CSI) with linac is used as a primary treatment. 1,2 But the superlong target volume poses some technical challenges include beam matching, junctions, and gaps of fields that lead to dose heterogeneity in the junction area. 3 A technique for overcoming these challenges is to shift the field boundaries weekly. 4 Other techniques include extending the source-to-skin distance, 5 the interactive movement of the couch, 6 field-in-field, 7 a linear ramp-like dose profile, 8 jagged-junction IMRT 9 , and split-field IMRT (sfIMRT), 3,10 and so on. The technique of sfIMRT with linac has become a major strategy for optimizing the whole craniospinal target volume simultaneously that offers several advantages, such as easier formulation of the radiotherapy plan, easier setup, reduced delivery time, more homogeneous dose, and superior sparing of organs at risk. 11 The ExacTrac system (BrainLAB AG, Feldkirchen, Germany) is a patient positioning system consisting of a radiographic kV x-ray imaging system for verifying patient position and a six degree-offreedom (6D) robotic couch for correcting patient position in six-dimensional directions. 12,13 ExacTrac offers several clinical benefits including faster patient alignment using the 6D robotic couch, the ability to monitor patient motion, and a reduction in image-based radiation delivered to the patient, 14 but it cannot provide much information due to a limited view of projections. In contrast, CBCT (cone-beam CT) is favored because it offers a three-dimensional view with better visualized anatomical structures and soft tissues than two-dimension (2D) imaging options. However, its applications are limited by relatively long image acquisition time, relatively high radiation to the patient, and other technical limitations. 15 An ideal IGRT method will minimize the dose without compromising the image guidance accuracy to prevent long-term side effects by reducing the integral dose is highly important in pediatrics due to the long life expectancy of the patients. The accuracy of the ExacTrac system is reported, and the ExacTrac represents an alternative to CBCT for CSI. 12,13 Experiments have been done for CSI in our previous study, 16 which prove that the accuracy of the ExacTrac system is consistent with the CBCT. The quality assurance of the infrared and the x-ray system is done on a weekly and daily frequency, respectively. The weekly check includes the radiation isocenter defined by a Winston-Lutz test, the isocenter calibration aligning the couch top with the linac isocenter, and the x-ray calibration calibrating with an ET (ExacTrac) X-ray Calibration Phantom. The daily check verifies the ExacTrac isocenter and x-ray calibration with a tungsten sphere located in the center of the ET Isocenter Phantom.
With the introduction of image guidance, positional variation can be measured and corrected in protocols. 17 In this study, the technique of sfIMRT with linac and the ExacTrac system were used to observe daily setup, intrafraction motion, and positional errors on the junctions. Besides, despite great effort to achieve precise repositioning and immobilization of patients and image guidance, radiation delivery uncertainties still exist due to junctions, residual setup errors, and intrafractional involuntary variations during verification procedure. Thus, the quantitative dosimetric effects of positional uncertainties also need to be very well understood to ensure the delivery of high-quality treatment to the CSI.

2.A | Patients
We report the outcomes of 15 unselected patients who were treated in our department between Oct. 2016 and Jan. 2018 with pathological detection, of whom 13 patients had germinomas, and two had medulloblastomas. The median age for these patients was 20 years (range 11-32 yr). All patients underwent tumor resection or biopsy. The pertinent information was tabulated (Table I).

2.B | . CT simulation and treatment planning
The patients were immobilized in a head-neck-shoulder thermoplastic mask with their arms resting at their sides, while a tattoo line along the longitudinal axis of the body was drawn to facilitate setup.
The GTV (Gross Tumor Volume) of the primary tumor and metastatic lesions was determined, and the CTV (Clinical Target Volume) included the whole brain, spinal cord, and terminal cisternae. The  (Table 1). All planning target volumes were optimized in synchrony using a commercial TPS (Eclipse version 8.9; Varian Medical Systems, Palo Alto, CA).
The iso1 was for cranial PCTV, iso2 was for the upper spinal cord PCTV, and iso3 was for the lower spinal cord PCTV. The field set iso1 consisted of five or seven fields with an average gantry angle, and the field sets iso2 and iso3 consisted of fields with three angles: 240°, 120°, and 180°.

2.C | Radiation therapy guided by the ExacTrac system
The ExacTrac system is based on a Varian Trilogy linear accelerator (Varian Medical Systems), which adopts 6-MV photon and a slidingwindow technique. The accelerator is equipped with a high-definition multileaf collimator (HD120 MLC) containing 120 leaves, 64 2.5 mm central leaves, and 56 5.0 mm peripheral leaves. The ExacTrac system was used for positioning in six directions (Lat (lateral), Lng (longitudinal), Vrt (vertical), Pitch, Roll, and Yaw and this order was followed throughout the article) and the flowchart of the workflow from patient setup to treatment was shown in Fig. 1. Firstly, the patient was positioned on the 6DoF couch and the couch was in the zero position at this moment, which meant the pitch, roll, and yaw angle were all set to 0°. Afterward, the patient was prepositioned with the aid of the infrared system, and this step was referred to as an initial setup that was then verified by using the x-ray component of ExacTrac and matched to the corresponding reference digitally reconstructed radiograph (DRR) automatically by a corresponding matching algorithm. The verification of the initial patient setup was referred to as initial verification and the setup error resulting from this initial verification was referred to as initial setup error (E 0 ). If the initial setup error was outside the tolerance (±2mm, ±2°), the determined correction was applied with the aid of the 6DoF couch. Two-kV images with ExacTrac had been acquired again until the setup error (E 1 ) was within tolerance, and then it was applied to fields. The position error was acquired similarly for the midtreatment ("mid-treatment" meant when half of all treatment fields were delivered) and posttreatment recorded as E 2 and E 3 . In the next step, the couch was moved from the first treatment isocenter to the second. Once the couch was moved, the new patient position was verified with the aid of the x-ray system of ExacTrac. Again, if the resulting moving couch error was outside the tolerance, the corresponding correction was applied and again verified until the error was within tolerance (E 4 ). The position error was also acquired for the midtreatment and posttreatment recorded as E 5 and E 6. In the third step, the second procedure was repeated for the third segment target volume and recorded position error as E 7 , E 8 , E 9 . E 0 was used to calculate the initial setup error. The postcorrection ExacTrac scan E 1 or precorrection scan E 0 , where the initial setup was within tolerance, was used to calculate the residual setup error. Nine/six ExacTrac scans (E 1 to E 9 /E 1 to E 6 ) of three/two isocenters target volume were conducted to monitor the intrafraction error. The variability E m of the transposition of isocenters was obtained, and the overall position variability on the junctions was obtained by summing the mean position variability between adjacent targets.
The displacement of patients was related to the duration of radiation delivery, ordinal number of fraction (named i), and time points of image acquiring (named t j ). Therefore, r ij , defined with the displacement vector of the patient, would be obtained with Lat, Lng, Vrt, Pitch, Roll, Yaw j . The misalignment of associated beam between two consecutive images was estimated by the following equation: following equation:ɛ ij ¼ r i;jþ1 À r i;j and the systematic error∑ i for the fraction i can be obtained by (1) whileN f;i was the total number of images acquired in the fraction i.
The random errorσ i for the fraction i also could be estimated from These values enabled us to obtain the mean population setup and the population random errorσas WhereN s was the total number of fractions included in this study.
Finally, the population systematic error∑was defined as the standard deviation of the patient systematic error, as follows: T A B L E I Patient characteristics, prescribed radiation dose, and fraction size.
The mean population setup error M, the population random error σ, and the population systematic error Σ were calculated via Eqs. (3)-(5). [18][19][20][21] The overall values of Σ and σ were defined as the root-mean-square of the setup error, the intrafraction error, and the transposition error of the isocenters. The geometric formula 2 Σ AE 0:7σ that was defined by van Herk 19 was used to calculate the margins of the CTVs to PCTVs.

2.D | Dosimetric effects of the longitudinal direction
To simulate the dosimetric effects of residual positional uncertainties, the original plan was copied to each new plan and recalculated in TPS using the same beam configuration and same total monitor unit, with different simulated isocenter position. The roll and yaw were rotated by changing the gantry angles and couch angles, respectively, but the rotation of the pitch was not an easy task.
Some investigators 22 initial verification x-ray completion apply treatment fields E4 (7) obtain eventual error E6 (9) obtain middle error E5 (8) spinal cord PCTV F I G . 1. Flowchart of the workflow from setup to treatment completion using the ExacTrac system. volume coverage and the maximal dose to the spinal cord were compared with those of the original plan.

2.E | Follow-up
Follow-up began upon completion of the patients' radiotherapy.
Craniocerebral MRI and spinal cord MRI examinations of patients were conducted every 3 months during the first 6 months and every 6 months after 6 months of radiotherapy. The tumor size, recurrence, metastasis were monitored and compared. Side effects were evaluated by RTOG acute radiation injury grading standard and CTCAE3.0 standard.

| RESULTS
A total of 2682 ExacTrac images, namely, 678 precorrection images Only three patients were within tolerance in the translation direction, and seven patients in the rotation direction for all fractions.
The areas in blue and pink (Fig. 2) Data on the initial and residual errors of the spinal cord PCTVs (the second and third PCTVs) due to the transposition of the treatment isocenters are also summarized in Table II. The distributions of the initial errors were almost within ± 16 mm (SD 4.5 mm) in the direction of translation and ± 6°(SD 1.3°) in the rotation direction except for one fraction that the minimum value of the error of third PCTV in the Lat direction is −24.8 mm, while the corresponding residual errors were within ± 2 mm and ± 1.25°, respectively. The most significant error was detected in the Lat direction.
The Dispersal of the initial errors were more substantial in the direction of translation than in the rotation direction, and the third PCTV was more significant than the second PCTV in six directions.
The dispersion was more extensive in the Lat directions than the Lng We conclude that the position must be corrected after the daily setup and the transposition of the isocenters.

3.B | Intrafraction motion
The Therefore, the intrafraction motion in the lateral direction should be monitored throughout treatment in consideration of the immobilization approach used in our department.

3.C | Position variability in the junction areas
The data in Table IV reveal that         In terms of HI of PGTV and PCTV, 60.2% and 58.8% of fractions increased, 0.5% and 2.9% remained unchanged, and 39.3% and 38.2% decreased. The HI of PGTV showed a maximum increase of 108.7%, compared to 56.6% for PCTV, except for patient 5, for whom the maximum growth was 115.7%.

3.F | Follow-up
The study was continued until May 2019, and the median follow- One patient developed cerebral edema 17 months after radiotherapy, and the extent of cerebral edema decreased after treatment.
Radiation myelitis was not observed after the craniospinal radiotherapy.
Therefore, the patients did not exhibit any side-effects by the overall treatment during the follow-up period.

| DISCUSSION
The ExacTrac image-guided radiotherapy system provides a fast and effective method for monitoring the position of the patients receiving craniospinal irradiation, and the cumulative dose is about 0.5-1 cGy in this study.
In this study, the initial setup errors in more than 96% of the fractions were in the range of (±4 mm, ±4°) in six directions according to the result of Beltran, 11 namely, children who were not localized using CBCT had a setup uncertainty of as large as 4 mm.
According to Beltran, errors > 2°should be corrected due to the nonnegligible changes in the gEUD for critical structures or target volumes. Peng et al. 26 demonstrate that rotational setup errors < 3°h ave a very minimal influence on the dose distribution.
Physiological movements such as respiratory motion, peristaltic motion, and heartbeat increase the setup uncertainty 27 and may result in deviation of the irradiated volume and organs at risk. 28,29 The treatment times for CSI delivery are often approximately 20 min or longer, which is associated with a higher risk of positional variation. 30 In our study, the delivery also took at least 20 min for a three-isocenter plan. Iglesias et al. 31  Transposition of treatment isocenter(s) is a significant factor that affects the hot and cold dosage spots in junction regions. Seppala et al. 10 propose a dynamic sfIMRT technique in CSI with isocenters that overlap at least 4 cm with each other, and the main benefit is that the homogeneous dose distribution is insensitive to alignment errors. Fred et al. 9 propose a similar technique that uses a threeisocenter jagged-junction IMRT plan that involves the realization of overlap in the junction regions using multiple beam directions: three for the spinal junctions and ten for the CSI junction. According to Fred, that their approach reduces the susceptibility of the junction areas to mismatching and the formation of hot and cold spots, as an error in one beam direction will not be compounded but will be DUAN ET AL.  Abbreviations: F = the first isocenter, S = the second isocenter, T = the third isocenter. +: the introduced errors were in the same direction to the coordinate axis; −: the introduced errors were in the opposite direction to the coordinate axis; The introduced errors resulted in a 96.7% reduction in the min dose of the spinal cord target volume in one patient, although the mean doses fluctuated within 0.027%; hence, a small portion of the target volume was not exposed. Inhomogeneity of the dose due to position errors can result in under dosage and local relapses in the cribriform and neuraxis areas. 40